youtube-transcript.ai

Google I/O '26 Keynote

Watch with subtitles, summary & AI chat
Add the free Subkun extension — works directly on YouTube.
  • Watch
  • Subtitles
  • Summary
  • Ask AI
Try free →

Google I/O 2026 showcased significant advancements in AI, particularly with the Gemini family of models and agentic capabilities. The event highlighted how AI is being integrated into everyday products and services, from enhanced search experiences and creative tools to personal assistants and scientific research. Google emphasized its commitment to responsible AI development and its vision for AI as a tool to solve complex global challenges.

Full Transcript (Bilingual)

https://www.youtube.com/watch?v=wYSncx9zLIU
Translation: zh-CN

[12:05] Google I/O 2026

[35:12] Wow!
哇!

[35:13] Wow!
哇!

[35:15] Hello!
你好!

[35:19] Hello!
你好!

[35:19] Hello! Wow, good morning, everyone! What is up, Google I/O?
你好!大家早上好!谷歌I/O怎么样?

[35:26] I'm Valkyrae, and I'm so excited to be here with you guys today.
我是Valkyrae,我非常激动今天能和大家在这里。

[35:29] And I'm Jack, AKA courage.
我是Jack,又名courage。

[35:31] Shout-out to everyone tuning into the livestream worldwide.
向所有收看全球直播的观众致敬。

[35:34] We are YouTube creators, gamers and occasionally competitors.
我们是YouTube的创作者、游戏玩家,偶尔也是竞争对手。

[35:39] I wouldn't call it much of a competition, though.
不过,我倒不认为这是什么激烈的竞争。

[35:41] Okay, hold on.
好的,等等。

[35:42] So today, we're going to be playing a game together.
所以今天,我们要一起玩一个游戏。

[35:46] It's called Infinite Scaler.
它叫做Infinite Scaler。

[35:47] And you can scan that big code right there to pull it up, no download required.
你可以扫描那边的二维码来启动它,无需下载。

[35:53] Infinite Scaler is basically a massive collaborative experiment.
Infinite Scaler基本上是一个大规模的协作实验。

[35:57] That means over the next 20 minutes, we won't just be competing against each other, although I'll definitely be getting the highest score, but we'll actually be building the
这意味着在接下来的20分钟里,我们不仅仅是互相竞争,虽然我肯定会获得最高分,但我们实际上将要构建这个

[36:08] We'll actually be building the game together as we play it.
我们将一起玩游戏,一边玩一边构建游戏。

[36:11] The game itself is pretty straightforward.
游戏本身非常直接。

[36:12] You bounce on these little jump pads to climb a tower as fast as you can.
你可以在这些小跳板上弹跳,以最快的速度爬上塔楼。

[36:17] If you're too slow, the lights go out below you, and you fall.
如果你太慢,你下方的灯就会熄灭,然后你就会掉下去。

[36:22] But you may have noticed that every-level looks completely unique.
但你可能已经注意到,每个关卡看起来都完全独特。

[36:25] That's because every block of the tower is a world you guys generate using your creativity and some AI magic.
那是因为塔楼的每个区块都是一个世界,你们通过创造力和一些人工智能魔法来生成它。

[36:32] That means every time you play, it's completely unique.
这意味着你每次玩的时候,它都是完全独特的。

[36:37] You can create a cyber punk city or a forest full of capybara.
你可以创建一个赛博朋克城市或一个充满水豚的森林。

[36:44] Whatever you dream up just type it in as long as it's safe for work.
无论你梦想什么,只要对工作安全就输入进去。

[36:46] That's right.
没错。

[36:46] We don't want to get HR involved.
我们不想让人力资源部介入。

[36:50] Look we'll give you a quick peek at how this actually works.
看,我们会让你快速了解它是如何工作的。

[36:52] You literally just type in a prompt.
你只需要输入一个提示词。

[36:56] It can super simple because our friend Gemini will automatically give it a polish for you on the back end.
它可以非常简单,因为我们的朋友 Gemini 会在后端自动为你润色。

[37:01] Then it gets sent over to --
然后它会被发送到 --

[37:03] Nano Banana, which makes the background and foreground sprite sheaths, adds a little tech magic, takes the
Nano Banana,它会制作背景和前景精灵鞘,添加一些技术魔法,获取

[37:08] a little tech magic, takes the depth map and boom you've got this awesome 3D effect.
一点点技术魔法,获取深度图,然后砰,你就有了这个很棒的 3D 效果。

[37:11] Like check out the one I made, a tribute to my Shiba Inu Mika.
就像看看我做的那个,献给我柴犬 Mika 的致敬。

[37:15] tribute to my Shiba Inu Mika.
献给我柴犬 Mika 的致敬。

[37:17] Look how cute he is.
看看他多可爱。

[37:17] So cute.
太可爱了。

[37:18] It's decent. Mine is way better.
还不错。我的好多了。

[37:23] Spaghetti slippery slide.
意大利面滑滑梯。

[37:24] Of course, of course.
当然,当然。

[37:25] All right, enough talking, let's play.
好了,说够了,我们来玩吧。

[37:27] Are you guys ready?
你们准备好了吗?

[37:27] You can jump right in on your phone or just grab the link on your computer.
你可以直接用手机加入,或者在电脑上获取链接。

[37:31] This is a play at your own pace type of game so don't stress if you're still getting your devices connected.
这是一个可以按自己节奏玩的游戏,所以如果你还在连接设备,别担心。

[37:36] There will be plenty of chances to jump in later.
之后将有很多机会加入。

[37:39] later.
稍后。

[37:39] All right.
好的。

[37:40] Let's go, let's get started.
我们开始吧,开始吧。

[37:42] Let's do this.
我们开始吧。

[37:44] Look, Rae no pressure, there's just thousands of people here in person and hundreds of thousands of people tuned in online.
看,Rae,没有压力,这里有成千上万的现场观众,还有数十万在线观众。

[37:49] Okay, I'm feeling the pressure, Jack.
好的,我感到压力了,Jack。

[37:51] All right.
好的。

[37:51] Let's go. Rae is going to get us started.
我们开始吧。Rae 将为我们开始。

[37:53] Again, you all with that QR code can join us with the link if you're watching online, from anywhere across the world.
再次,你们所有持有二维码的人都可以通过链接加入我们,如果你是在线观看,无论是在世界的哪个角落。

[38:00] Rae is going to show you how it's done.
Rae 将向你展示如何操作。

[38:02] Let's see what record you can set.
让我们看看你能创造什么记录。

[38:04] Okay. It's pretty simple.
好的。这很简单。

[38:04] You just have to hit the jump pads because there is no jump button.
你只需要踩跳跃垫,因为没有跳跃按钮。

[38:07] And each room
每个房间

[38:11] No jump button.
没有跳跃按钮。

[38:15] And each room is just so freaking unique and is just so freaking unique and random.
而且每个房间都非常独特,而且非常独特和随机。

[38:20] I have to lock in.
我必须锁定。

[38:21] A hand, these are prompts from you guys at home,
一只手,这些是你们在家里的提示,

[38:23] that you're sending on in and
你们发过来的,而且

[38:26] we're also, as we play, midway through going to take a look at
我们也会,在我们玩的时候,中途会看看

[38:28] which players set the record, as well as which countries have the
哪些玩家创造了记录,以及哪些国家拥有

[38:31] most levels and it's important to get up the right side, the rocket.
最多的关卡,而且重要的是要走右边,火箭。

[38:38] Rocket.
火箭。

[38:39] You need the rocket.
你需要火箭。

[38:43] I'm rooting for the Philippines.
我支持菲律宾。

[38:43] Who are you rooting for, Jack?
你支持谁,杰克?

[38:46] You think the Philippines are going to pop up a?
你认为菲律宾会冒出来吗?

[38:48] I think Japan.
我认为是日本。

[38:50] Okay, Japan.
好的,日本。

[38:50] Japan is going to do great.
日本会做得很好。

[38:53] Maybe Brazil. All right, so far, it's 30.
也许是巴西。好了,到目前为止,是30。

[38:56] Look if anyone in the crowd can get over 50 I'll say you're pretty darn good.
看看人群中是否有人能超过50,我会说你很棒。

[38:59] We did the a lot of tests and warmups, and hit some pretty crazy records.
我们做了很多测试和热身,并创造了一些非常疯狂的记录。

[39:04] Here it comes.
来了。

[39:10] Jack, you're stressing me out.
杰克,你让我很紧张。

[39:13] Stressing me out.
这让我压力很大。

[39:13] Basketball uh-oh.
篮球,哦不。

[39:14] Basketball uh-oh.
篮球,哦不。

[39:14] Uh-oh.
哦不。

[39:19] So good.
太好了。

[39:19] Okay.
好的。

[39:21] So good.
太好了。

[39:21] Okay.
好的。

[39:23] 43 levels, new points record.
43级,新的得分记录。

[39:23] Pretty good.
相当不错。

[39:25] Points record.
得分记录。

[39:25] Pretty good.
相当不错。

[39:28] Let me do it one more time.
让我再做一次。

[39:31] Okay.
好的。

[39:31] My goal is at least 50.
我的目标是至少50。

[39:33] Okay.
好的。

[39:33] My goal is at least 50.
我的目标是至少50。

[39:34] What do you think the most prompted things are going to be?
你认为最常被提示的东西会是什么?

[39:35] I feel like food, mainly, because they have really good food here.
我觉得是食物,主要是因为这里的食物真的很好。

[39:37] Mainly, because they have really good food here.
主要是因为这里的食物真的很好。

[39:37] Good food here.
这里的美味食物。

[39:38] I think people are going to prompt in their pets.
我认为人们会提示他们的宠物。

[39:39] Going to prompt in their pets.
会提示他们的宠物。

[39:40] Who doesn't want to see their pets?
谁不想看到他们的宠物呢?

[39:42] Pets?
宠物?

[39:43] That's like a fork battle going on.
这就像一场叉子大战。

[39:46] Battle going on.
正在进行的战斗。

[39:46] Robots controlling a ship.
机器人控制着一艘船。

[39:48] Controlling a ship.
控制着一艘船。

[39:49] I want to just look at each room, but they're going by so fast.
我想看看每个房间,但它们闪过得太快了。

[39:52] At each room, but they're going by so fast.
看看每个房间,但它们闪过得太快了。

[39:55] By so fast.
闪过得太快了。

[39:58] There's a capybara.
有一只水豚。

[39:58] Capybara.
水豚。

[39:58] Masseuse.
按摩师。

[39:58] Funny.
有趣。

[39:59] Masseuse.
按摩师。

[39:59] Funny.
有趣。

[39:59] It's pretty therapeutic, honestly.
说实话,这相当有疗愈作用。

[40:03] It's pretty therapeutic, honestly.
说实话,这相当有疗愈作用。

[40:03] Honestly.
说实话。

[40:04] Maybe jellyfish DJ's would be a good one.
也许水母DJ会是个好主意。

[40:06] DJ's would be a good one.
DJ会是个好主意。

[40:06] That's what we just saw.
那正是我们刚才看到的。

[40:07] That's what we just saw.
那正是我们刚才看到的。

[40:08] I can't believe they introd for us.
我不敢相信他们为我们介绍了。

[40:08] A cow in
一头牛在

[40:15] They introduced for us. A cow in space.
他们为我们介绍了。一头太空中的牛。

[40:15] Space.
太空。

[40:15] That's awesome.
太棒了。

[40:16] That's awesome.
太棒了。

[40:16] And again, you all can join worldwide right now, no download required, a link at the top of your screen.
再说一次,你们现在都可以加入,加入全球,无需下载,屏幕顶部有链接。

[40:22] All right. You're at 31.
好的。你得了31分。

[40:25] At 31.
得了31分。

[40:27] Okay. I'm aiming for 50.
好的。我的目标是50分。

[40:30] What do you think the high score is going to be?
你认为最高分会是多少?

[40:32] Jack, what do you think the high score is going to be?
杰克,你认为最高分会是多少?

[40:33] Going to be?
会是多少?

[40:36] I think the high score, someone out there is going to get 100 plus.
我认为最高分,有人会拿到100多分。

[40:37] It's going to be crazy.
这将会是疯狂的。

[40:40] Uh-oh, you got it. You got it. You're good.
哦哦,你做到了。你做到了。你很棒。

[40:43] You're stressing me out.
你让我很紧张。

[40:45] The thing to know about this is that for each room, you have to hit the jump pad to get through the hole.
关于这个要知道的是,每个房间你都必须踩到跳板才能穿过洞口。

[40:51] You can't just go through the bottom.
你不能直接从底部穿过。

[40:56] Are you going to have tips and tricks?
你会有一些技巧和窍门吗?

[40:59] I have not had the chance to go.
我还没有机会去。

[41:01] We're good. There we go. Okay.
我们做得很好。就这样。好的。

[41:04] Don't worry, I know the crowd is going to get real loud when it's my turn.
别担心,我知道轮到我的时候,观众会非常吵。

[41:06] I'll give someone $5, please.
我给某人5美元,拜托了。

[41:11] Okay.
好的。

[41:12] You can't bribe the crowd.
你不能贿赂观众。

[41:13] That's cheating.
那是作弊。

[41:14] I've got tricks up my sleeve.
我还有绝招。

[41:19] I've got tricks up my sleeve I'm going to hope.
我还有绝招,我会尽力而为。

[41:20] I'm going to hope.
我会尽力而为。

[41:22] That was so cute.
那太可爱了。

[41:22] That was so cute.
那太可爱了。

[41:24] Okay. No pressure.
好的。没有压力。

[41:24] Again, just literally competing with everybody.
再说一遍,只是字面意义上的与大家竞争。

[41:29] competing with everybody.
与大家竞争。

[41:29] Jack.
杰克。

[41:29] Jack.
杰克。

[41:30] Only millions of people.
只有数百万人。

[41:30] people.
人。

[41:32] Jack... Okay. I'm going to be yapping your ear off when you do it.
杰克……好的。当你这样做的时候,我会喋喋不休地跟你说。

[41:36] going to be yapping your ear off when you do it.
喋喋不休地跟你说。

[41:36] I think it's really good.
我认为这真的很好。

[41:37] really good.
真的很好。

[41:38] That's crazy.
太疯狂了。

[41:38] That's crazy.
太疯狂了。

[41:39] Now, it's my turn, everybody! Let's go.
现在,轮到我了,大家!我们开始吧。

[41:41] Let me show you how it's done.
让我向你展示它是如何完成的。

[41:43] show you how it's done.
展示它是如何完成的。

[41:46] Show you how it's done.
展示它是如何完成的。

[41:49] Infinite Scaler, live at Google I/O!
无限扩展器,在谷歌 I/O 大会上直播!

[41:52] Starts now M.
现在开始 M。

[41:53] It's not too late to join in.
现在加入还不晚。

[41:55] We're going to look at the high scores right after this, or maybe after a couple of rounds, because who knows how far Jack is going to get.
我们将在之后立即查看高分,或者可能在几轮之后,因为谁知道杰克能走多远。

[41:58] this, or maybe after a couple of rounds, because who knows how far Jack is going to get.
之后,或者可能在几轮之后,因为谁知道杰克能走多远。

[42:03] far Jack is going to get.
杰克能走多远。

[42:05] Good. Rocket ship already.
好。火箭已经升空。

[42:06] I'm locked in.
我已锁定。

[42:08] locked in.
锁定。

[42:09] If you're at home on your PC, scan the QR code.
如果你在家用电脑,请扫描二维码。

[42:13] on your PC, scan the QR code.
用你的电脑,扫描二维码。

[42:13] Double rockets?
双火箭?

[42:15] Yep, I rigged it.
是的,我作弊了。

[42:15] That's crazy.
太疯狂了。

[42:20] That's crazy.
太疯狂了。

[42:20] Really?
真的吗?

[42:20] Really?
真的吗?

[42:21] Let's see how far you can get.
让我们看看你能走多远。

[42:25] you can get.
你能走多远。

[42:26] The bears.
熊。

[42:26] The bears.
熊。

[42:27] Oh, my gosh.
哦,我的天哪。

[42:27] Oh, my gosh.
哦,我的天哪。

[42:29] What is going on?
这是怎么回事?

[42:29] What is going on?
这是怎么回事?

[42:30] All the rockets, this is so rigged.
所有的火箭,这太不公平了。

[42:33] This is not fair right now.
现在这不公平。

[42:38] now. Candy canes.
现在。拐杖糖。

[42:39] Candy canes.
拐杖糖。

[42:40] I'm going for 100.
我要去100。

[42:40] I'm going for 100.
我要去100。

[42:41] Who did you actually pay?
你到底付了多少钱?

[42:42] actually pay?
你付了多少钱?

[42:42] Don't worry about it.
别担心。

[42:43] it. That was close.
它。差点就成功了。

[42:46] That was close. The bears and the river.
差点就成功了。熊和河。

[42:47] the river.
河。

[42:48] Hot-air balloons.
热气球。

[42:48] Hot-air balloons.
热气球。

[42:49] I like watching you play, because I can actually look at the rooms.
我喜欢看你玩,因为我可以看到房间。

[42:50] you play, because I can actually look at the rooms.
你玩,因为我可以看到房间。

[42:53] look at the rooms.
看房间。

[42:54] There's another.
还有另一个。

[42:54] There's another.
还有另一个。

[42:56] There we go. More food, please.
好了。请再来点食物。

[42:59] food, please.
食物,拜托了。

[43:00] Puppies. Washing machines with a ship in it.
小狗。里面有船的洗衣机。

[43:07] machines with a ship in it. You're at 70.
里面有船的机器。你现在70了。

[43:08] You're at 70.
你现在70了。

[43:09] What did you get to six? I'm about to pass that right now.
你得了多少分到六?我马上就要超过了。

[43:11] to six? I'm about to pass that right now.
到六?我马上就要超过了。

[43:12] right now. I'm about to pass that right now.
现在。我马上就要超过了。

[43:13] that right now.
现在。

[43:14] You have to try to get 100.
你必须努力拿到100。

[43:14] get 100.
拿到100。

[43:17] I will. Don't worry.
我会的。别担心。

[43:18] Don't worry. If you're playing at home when you look at the leaderboard, it's for you all at
别担心。如果你在家玩,当你看到排行榜时,它是给你们所有人的

[43:19] home when you look at the leaderboard, it's for you all at
家,当你看到排行榜时,它是给你们所有人的

[43:21] Leaderboard, it's for you all at home.
排行榜,是给你们所有在家的人的。

[43:21] I won't be on that.
我不会在上面。

[43:26] I've had a lot of practice...
我练习了很多次...

[43:28] So I may or may not have been here in this seat for two hours yesterday playing.
所以我可能在这里坐了两个小时,也可能没有,昨天在玩。

[43:33] We did get to practice.
我们确实得到了练习。

[43:38] This is actually getting nuts now.
这真的开始变得疯狂了。

[43:39] You're already almost at 100.
你已经快到100了。

[43:42] I wonder which country actually will wind up getting the top.
我想知道哪个国家最终会拿到第一名。

[43:46] Congratulations.
恭喜。

[43:46] You did it.
你做到了。

[43:51] Jack... It's getting dark.
杰克……天快黑了。

[43:56] Yes you better hurry up.
是的,你最好快点。

[43:57] It's fine.
没关系。

[43:57] Everything's fine.
一切都好。

[44:01] A ship in a bottle?
瓶子里的船?

[44:03] Yeah. Pigeons.
是的。鸽子。

[44:06] Well, Rae.
嗯,瑞。

[44:10] Look at the bee!
看那只蜜蜂!

[44:11] I think that might be my favorite.
我想那可能是我的最爱。

[44:12] The bunnies with the boba.
带珍珠奶茶的兔子。

[44:16] Now, I want boba.
现在,我想喝珍珠奶茶。

[44:21] You're at 125, Jack.
你在125,杰克。

[44:21] That cheeseburger has lettuce and tomatoes.
那个芝士汉堡有生菜和番茄。

[44:25] Can we get more animals with clothes on?
我们能看到更多穿着衣服的动物吗?

[44:28] No burger should have -- let's just cancel the keynote, and I'll keep playing.
没有汉堡应该有——我们取消主题演讲吧,我继续玩。

[44:35] I'm sure.
我确定。

[44:35] I think that's what everyone here wants.
我想这正是这里每个人想要的。

[44:38] They would love that.
他们会喜欢那样的。

[44:41] You're at 130.
你在130。

[44:43] I wish you won't first.
我希望你不要先。

[44:43] This is crazy.
这太疯狂了。

[44:47] Okay.
好的。

[44:48] Okay. 142!
好的。142!

[44:50] That was pretty good.
那相当不错。

[44:50] Thank you.
谢谢你。

[44:51] That was pretty good.
那相当不错。

[44:52] Look at these, numbers.
看看这些,数字。

[44:58] It's over 157 countries.
它覆盖了157个国家。

[45:00] This is in the last 10 minutes.
这是在过去的10分钟里。

[45:03] 14,000 people!
14000人!

[45:03] That that's wild.
那真是太疯狂了。

[45:06] That's actually a lot of people.
那实际上是很多人。

[45:09] This just happened.
这刚刚发生。

[45:10] 1818 levels.
1818个关卡。

[45:11] Let's check out the top scores..
让我们来看看最高分。。

[45:16] Brawny icy moss with 248.
Brawny icy moss 获得248分。

[45:16] We suck.
我们太差了。

[45:19] How did they even have that much time?
他们怎么会有那么多时间?

[45:21] I don't know.
我不知道。

[45:23] VALKYRAE: I don't know. That's actually nuts.
VALKYRAE:我不知道。这太疯狂了。

[45:26] That's actually nuts.
这太疯狂了。

[45:27] COURAGEJD: Top country, United States.
COURAGEJD:顶级国家,美国。

[45:30] Shout-out India, 9,567 levels scaled, okay.
向印度致敬,9,567个关卡已缩放,好的。

[45:34] 9,567 levels scaled, okay.
9,567个关卡已缩放,好的。

[45:35] VALKYRAE: 33,000 points, that's crazy.
VALKYRAE:33,000分,太疯狂了。

[45:36] Then we've got Germany, Brazil, France, Mexico, Canada, Poland, Spain.
然后我们有德国、巴西、法国、墨西哥、加拿大、波兰、西班牙。

[45:40] Where's the Philippines?
菲律宾在哪里?

[45:46] COURAGEJD: Don't worry. Next one.
COURAGEJD:别担心。下一个。

[45:46] We'll take a look at some of the levels that people have made.
我们将看看人们制作的一些关卡。

[45:49] VALKYRAE: I'm very excited for this.
VALKYRAE:我对此非常兴奋。

[45:50] Let's see what you guys have made.
让我们看看你们都做了什么。

[45:53] Okay. We've got a duck with a sweater fishing, by poppy neon gravy.
好的。我们有一只穿着毛衣钓鱼的鸭子,作者是poppy neon gravy。

[45:59] gravy.
gravy。

[46:02] COURAGEJD: Very cool.
COURAGEJD:非常酷。

[46:03] VALKYRAE: Cute. Are those like --
VALKYRAE:可爱。那些是像——

[46:05] COURAGEJD: A Coral reef.
COURAGEJD:一个珊瑚礁。

[46:12] Little anenomes with eyeballs.
长着眼睛的小海葵。

[46:13] You've got to get rid of the lettuce and tomato.
你得去掉生菜和番茄。

[46:16] VALKYRAE: So you're a cheese and meat only kind of guy.
VALKYRAE:所以你是个只吃奶酪和肉的人。

[46:19] This is by smiley science.
这是smiley science的作品。

[46:24] is by smiley science.
是来自笑脸科学的。

[46:24] Now, I'm hungry.
现在,我饿了。

[46:26] That makes sense.
这说得通。

[46:27] We saw this one in the levels.
我们在关卡里看到了这个。

[46:30] That's a really cool one.
那是一个很酷的东西。

[46:31] Ship in a bottle with little people around it.
瓶子里的船,周围有小人。

[46:40] A disco snail.
一只迪斯科蜗牛。

[46:41] How do you even think of that?
你怎么会想到这个?

[46:43] Snail where the shell is a disco ball.
蜗牛的壳是一个迪斯科球。

[46:46] Giant rock hand, asteroid hand by plushy hiccup.
巨大的岩石手,小毛绒打嗝的手,小行星手。

[46:50] I feel like we've got to give you another chance, though.
我觉得我们得再给你一次机会。

[46:53] Me?
我?

[46:54] Yeah. You've got to get 100 plus.
是的。你得拿到100多分。

[46:57] Okay. I can hit 100.
好的。我可以达到100。

[47:00] People already hit 250...
人们已经达到了250...

[47:01] Yeah. Maybe we can redeem ourselves.
是的。也许我们可以救赎自己。

[47:05] Look at this crowd here, that's awesome.
看看这里的观众,太棒了。

[47:06] This is crazy.
这太疯狂了。

[47:07] Some of you all out there in the sun, make sure you put on some sun screen.
外面在阳光下的一些朋友们,一定要涂防晒霜。

[47:11] This crowd is going all the way back now.
现在观众已经排到后面去了。

[47:14] It's going to get hot today.
今天会很热。

[47:15] No pressure.
没有压力。

[47:17] Some of the most important people in the world in the room, but no pressure.
世界上一些最重要的人都在这里,但没有压力。

[47:23] No pressure. I'm scaling.
没有压力。我在爬升。

[47:25] Scaling.
缩放。

[47:26] You all at home can join up and play right now.
你们在家的人可以立即加入并一起玩。

[47:27] I'm going to try to challenge you to get that leader board.
我将尝试挑战你们,争取登上排行榜。

[47:31] You have to get 220, I think.
我想你们必须达到220分。

[47:34] The Phoenix is cool.
凤凰很酷。

[47:35] A bunch of crystals.
一堆水晶。

[47:39] I passed the rocket.
我通过了火箭。

[47:39] I passed the rocket.
我通过了火箭。

[47:41] I didn't need it.
我不需要它。

[47:44] What are you doing?
你在做什么?

[47:46] I'm freakin' out, Jack.
我快要疯了,杰克。

[47:47] Okay.
好的。

[47:49] Uh-oh.
糟糕。

[47:50] Okay, okay.
好的,好的。

[47:50] Stay calm.
保持冷静。

[47:56] This is spaghetti.
这是意大利面。

[47:58] Rocket. Big. Okay.
火箭。大的。好的。

[47:59] Now, you've got it.
现在,你明白了。

[48:00] Someone in the crowd right now is watching and probably already at like 500,
现在人群中有人在看,可能已经达到500分了,

[48:01] and they might play for the whole show.
他们可能会玩一整场。

[48:08] I'm kind of curious what the top score is now.
我有点好奇现在的最高分是多少。

[48:14] You better make a mistake, people are going to be on their phones while they're doing it.
你最好别犯错,否则人们在做的时候会玩手机。

[48:17] Boing, boing, boing.
叮,叮,叮。

[48:18] Big rocket.
大火箭。

[48:21] This is the Jack luck I was looking for.
这就是我想要的杰克运气。

[48:25] You know, we got
你知道,我们得到了

[48:26] You know, we got told back stage that people here in the crowd today are from countries all around the world.
你知道,我们在后台被告知,今天这里的观众来自世界各地。

[48:31] Where did you guys come in from for today's event?
你们今天是从哪里来参加这次活动的?

[48:36] Can some people yell where you're from?
有人能喊出你们来自哪里吗?

[48:38] I heard Canada.
我听到了加拿大。

[48:39] I just heard like 40 different countries at once.
我刚才同时听到了大约40个不同的国家。

[48:42] And it all blended into one.
它们都融为一体了。

[48:45] But enjoy the nice weather.
但请享受这美好的天气。

[48:48] Boing, boing, boing... This is like the best head-empty type of game.
砰,砰,砰……这就像是最好的那种什么都不想的放松游戏。

[48:57] You broke your previous record.
你打破了你之前的记录。

[49:00] That's all I wanted.
这就是我想要的。

[49:00] I wanted to do better.
我想做得更好。

[49:03] There's a rocket to the left.
左边有个火箭。

[49:05] Oh, my gosh so clutch, dude.
哦,我的天,太关键了,哥们。

[49:09] I like how the crowd reacts to you getting that last one.
我喜欢观众们对你拿到最后一个时的反应。

[49:10] Gur, you're good.
古尔,你很棒。

[49:13] Let's go.
冲啊。

[49:17] It's fun seriously.
说真的,很有趣。

[49:20] This is your last chance.
这是你的最后一次机会了。

[49:21] It's over for me.
对我来说结束了。

[49:21] It's over for me.
对我来说结束了。

[49:21] I didn't hit 100.
我没有达到100。

[49:25] If I don't get on
如果我没有上去

[49:26] If I don't get on the global leader board then the global leader board then Google I/O is canceled.
如果我不能登上全球排行榜,那么全球排行榜,那么谷歌I/O就取消了。

[49:30] Here we go.
我们开始吧。

[49:35] The Google executive back stage is like tackle this guy off the stage right now.
谷歌高管在后台就像要把这个人从舞台上推下去一样。

[49:38] I'm going to have to.
我不得不这样做。

[49:41] I'm going to wave my hand in front of your screen.
我将在你的屏幕前挥手。

[49:42] Lock in, Jack.
集中注意力,杰克。

[49:43] That one is so cute.
那个真可爱。

[49:50] Cheese!
奶酪!

[49:52] Who doesn't love cheese.
谁不爱奶酪呢。

[49:54] The lactose intolerant.
乳糖不耐受者。

[49:57] We're good.
我们很好。

[49:59] Rocket.
火箭。

[50:00] Great from the start.
开局很棒。

[50:03] Again, guys, take a look at that link, join on in, rep your country, see if you can get a high score.
再次,各位,看看那个链接,加入进来,代表你的国家,看看你是否能获得高分。

[50:10] No download required, which was awesome to hear, because I feel like that's the hardest part.
无需下载,这听起来很棒,因为我觉得这是最难的部分。

[50:11] Taking the time to download the game, but you don't have to download.
花时间下载游戏,但你不需要下载。

[50:13] If you're at home, scan the QR code.
如果你在家,扫描二维码。

[50:20] Is that actually how it works?
它真的像这样运作吗?

[50:21] I wassed a hiring the horse wearing clothes.
我被雇佣了,那匹马穿着衣服。

[50:25] Yeah.
是的。

[50:26] You can hit
你可以按

[50:26] Yeah. You can hit the link in the top corner.
是的。你可以点击顶部的链接。

[50:28] the link in the top corner. Check it out.
顶部的链接。看看它。

[50:30] Check it out.
看看它。

[50:33] I'm in my flow state right now.
我现在正处于我的状态。

[50:35] I feel like you've had a lot of practice.
我觉得你练习了很多。

[50:39] I'm chilling right now.
我现在很放松。

[50:43] Bunnies with pujams.
穿着睡衣的兔子。

[50:45] Oh, my gosh, wolves chasing the little bees.
哦,我的天哪,狼在追逐小蜜蜂。

[50:52] You're at 57 right now.
你现在是57。

[50:55] The light is coming back.
光线正在回来。

[50:58] Nice. That's the crowd right there.
好的。那就是那群人。

[51:03] It's my good luck charm.
这是我的幸运符。

[51:06] Dragon. Is that your spaghetti slide?
龙。那是你的意大利面滑梯吗?

[51:09] That might have been it.
那可能就是了。

[51:13] Was that your room?
那是你的房间吗?

[51:17] Good call.
说得好。

[51:19] I haven't seen Mika yet. Less spiders, please.
我还没见到米卡。请少来点蜘蛛。

[51:23] Almost there. At 87. Oh, no. It's a disaster.
快到了。在87。哦,不。这是一场灾难。

[51:26] You're good, you're good.
你很好,你很好。

[51:30] Good.
好的。

[51:31] My last chance, I've got to lock in.
这是我最后的机会了,我必须锁定。

[51:33] I've got to lock in.
我必须锁定。

[51:35] You're almost to 100.
你快到100了。

[51:40] Congratulations, you did it.
恭喜你,你做到了。

[51:48] I'm really happy for you.
我真为你高兴。

[51:49] We're good.
我们很好。

[51:59] It's fine.
没关系。

[52:01] They're going to kick us out, Rae.
他们要把我们踢出去,Rae。

[52:09] Jack, you're way too good.
Jack,你太厉害了。

[52:13] How much have you practiced?
你练习了多久?

[52:17] I may have loaded this up at home.
我可能在家加载了这个。

[52:17] It's fine.
没关系。

[52:19] Spaghetti tower.
意大利面塔。

[52:21] People love spaghetti.
人们喜欢意大利面。

[52:23] The dogs are great.
狗狗们很棒。

[52:27] So much spaghetti.
这么多意大利面。

[52:28] You're at 140.
你现在是140。

[52:28] Oh, my gosh.
哦,我的天哪。

[52:34] Oh, my gosh. This might be the most people I've ever played video games in front of.
哦,我的天哪。这可能是我玩电子游戏以来在最多人面前玩的一次了。

[52:39] You could beat the crowd if you hit 250, unless someone else has beat it by now.
如果你达到250,你就能战胜观众,除非现在已经有人打破了这个记录。

[52:45] If I hit the world record, I become the Google CEO, that's how it works.
如果我打破世界纪录,我就成为谷歌CEO,事情就是这样运作的。

[52:49] I won't let that happen. Don't worry, Mr. CEO, I will tackle this man.
我不会让那种事情发生的。别担心,CEO先生,我会制服这个男人。

[52:58] Don't worry. Oh, Rae!
别担心。哦,Rae!

[52:59] I think you got this.
我觉得你能行。

[53:04] Rae, please!
Rae,拜托!

[53:06] You're at 180.
你现在是180。

[53:09] Okay. Hold on.
好的。等等。

[53:11] But I think it was 250 something. But I'm pretty sure they're still going.
但我想是250多。但我很确定他们还在继续。

[53:13] I have a feeling the high score is going to be double.
我有一种预感,高分将会翻倍。

[53:19] Maybe we'll take another look in a moment and find out.
也许我们稍后会再看一眼,然后弄清楚。

[53:25] There's my favorite, my fishing duck.
这是我最喜欢的,我的钓鱼鸭。

[53:28] You're at 200. And you have a rocket.
你现在是200。你还有一个火箭。

[53:33] Good! What more do I need?
好!我还需要什么?

[53:35] I need?
我需要?

[53:35] You need at least 50.
你至少需要50个。

[53:37] 50?
50个?

[53:42] Yeah, at least.
是的,至少。

[53:43] Nice.
不错。

[53:45] Rae, Rae. This is the best run I've ever had.
Rae,Rae。这是我玩过的最好的一次。

[53:47] And I did it on stage!
而且我是在舞台上做到的!

[53:55] This is like our jobs.
这就像我们的工作一样。

[53:56] They brought me here to make you look good.
他们把我带来是让你看起来很棒。

[54:05] Live streaming.
直播。

[54:05] Okay.
好的。

[54:07] Jack, you're at 236.
Jack,你现在是236。

[54:13] Keep going.
继续。

[54:13] Yo everyone at home!
嘿,家里的各位!

[54:15] You're lucky I can't get on that leaderboard, except for you guys.
你们很幸运,我无法登上那个排行榜,除了你们这些家伙。

[54:20] I'm actually so curious.
我实际上非常好奇。

[54:21] Let's go! Thank you, thank you!
加油!谢谢你,谢谢你!

[54:24] It was the 600 milligrams of caffeine.
这是600毫克的咖啡因。

[54:28] 250, Rae look, you know what? I got it.
250,Rae看,你知道吗?我明白了。

[54:30] I'm going to go.
我要走了。

[54:34] Least get the last rocket.
至少拿到最后一个火箭。

[54:34] No, no. And we'll
不,不。我们将会

[54:36] >> COURAGEJD: No, no. And we'll take it. Thank you, thank you.

[54:39] take it. Thank you, thank you. That was fun. Wait!

[54:42] That was fun. Wait! >> VALKYRAE: 20,000 of you have

[54:43] >> VALKYRAE: 20,000 of you have joined in! That's so many

[54:47] joined in! That's so many people.

[54:47] people. >> COURAGEJD: 163 countries.

[54:48] >> COURAGEJD: 163 countries. >> VALKYRAE: Do you think the

[54:49] >> VALKYRAE: Do you think the leader board has changed?

[54:51] leader board has changed? >> COURAGEJD: Let's take a look

[54:52] >> COURAGEJD: Let's take a look at the leader board.

[54:54] at the leader board. >> VALKYRAE: 259, let's see if

[54:55] >> VALKYRAE: 259, let's see if they've beaten your score.

[55:02] they've beaten your score. >> COURAGEJD: What? 800 --

[55:03] >> COURAGEJD: What? 800 -- >> VALKYRAE: How did you have

[55:03] >> VALKYRAE: How did you have the time? We just started.

[55:05] the time? We just started. >> COURAGEJD: No, someone check

[55:05] >> COURAGEJD: No, someone check that person's phone.

[55:09] that person's phone. >> VALKYRAE: Okay.

[55:09] >> VALKYRAE: Okay. >> COURAGEJD: Maybe that's just

[55:11] >> COURAGEJD: Maybe that's just Gemini playing.

[55:12] Gemini playing. >> VALKYRAE: Recent gleamy

[55:13] >> VALKYRAE: Recent gleamy wombat from the United States.

[55:14] wombat from the United States. Congratulations being number one

[55:15] Congratulations being number one right now.

[55:16] right now. >> COURAGEJD: And let's --

[55:19] >> COURAGEJD: And let's -- >> VALKYRAE: Top countries. Of

[55:20] >> VALKYRAE: Top countries. Of course, the United States.

[55:21] course, the United States. >> COURAGEJD: Wow. Okay.

[55:23] >> COURAGEJD: Wow. Okay. 74,000 points. That's insane.

[55:27] 74,000 points. That's insane. India, United Kingdom, Germany,

[55:30] India, United Kingdom, Germany, Brazil. So cool. France,

[55:31] Brazil. So cool. France, Canada, Mexico, Poland and Spain

[55:33] Canada, Mexico, Poland and Spain round out our top 10.

[55:35] round out our top 10. >> VALKYRAE: I can't believe how

[55:37] >> VALKYRAE: I can't believe how many people jumped in. That was

[55:39] many people jumped in. That was amazing.

[55:39] amazing. >> COURAGEJD: Look guys, that is

[55:40] >> COURAGEJD: Look guys, that is it from us. Thank you so much.

[55:41] it from us. Thank you so much. That was a lot of fun.

[55:43] That was a lot of fun. >> VALKYRAE: That was awesome!

[55:44] >> VALKYRAE: That was awesome! Have a great day you guys! And

[55:46] Have a great day you guys! And enjoy the show.

[55:48] enjoy the show. >> COURAGEJD: Enjoy the rest of

[55:49] >> COURAGEJD: Enjoy the rest of I/O. Bye, everyone.

[55:51] I/O. Bye, everyone. >> VALKYRAE: Bye.

[55:52] >> VALKYRAE: Bye. ♪

[55:53] ♪ ♪

[01:04:58] ♪ >> Take a second to realize how

[01:04:59] >> Take a second to realize how far we've come.

[01:05:05] far we've come. Empowered by all the things

[01:05:07] Empowered by all the things we've made together.

[01:05:08] we've made together. >> How many people have looked

[01:05:09] >> How many people have looked at a problem and then have been

[01:05:11] at a problem and then have been bold enough to say, I can solve

[01:05:12] bold enough to say, I can solve it?

[01:05:20] it? >> And staring at a future.

[01:05:21] >> And staring at a future. We all get to decide.

[01:05:21] We all get to decide. >> AI, as the ultimate tool to

[01:05:22] >> AI, as the ultimate tool to solve all the world's most

[01:05:23] solve all the world's most complex scientific problems.

[01:05:26] complex scientific problems. >> We're taking flight.

[01:05:27] >> We're taking flight. Every one of us is packed with

[01:05:31] Every one of us is packed with potential, the potential to mak

[01:05:33] potential, the potential to mak anything we can dream of.

[01:05:36] anything we can dream of. >> I do not know personally how

[01:05:38] >> I do not know personally how to code, but given the option t

[01:05:40] to code, but given the option t create something just opens up

[01:05:42] create something just opens up so many doors for me as an

[01:05:42] so many doors for me as an educator.

[01:05:44] educator. >> I discovered Gemini could

[01:05:45] >> I discovered Gemini could help me with a lot of things

[01:05:47] help me with a lot of things from menus to marketing and

[01:05:49] from menus to marketing and inventory.

[01:05:50] inventory. >> But in a world where anythin

[01:05:52] >> But in a world where anythin is possible.

[01:05:53] is possible. >> We're really pushing at this

[01:05:55] >> We're really pushing at this big mission, the mission of

[01:05:55] big mission, the mission of solvable disease.

[01:05:56] solvable disease. This is huge, but we have a

[01:05:58] This is huge, but we have a profoundly new technology in ou

[01:05:59] profoundly new technology in ou hands.

[01:06:00] hands. >> What matters is what we

[01:06:02] >> What matters is what we choose to build.

[01:06:04] choose to build. >> Hi, my name is Matteo.

[01:06:07] >> Hi, my name is Matteo. My brother Tomas made me this A

[01:06:11] My brother Tomas made me this A assistant.

[01:06:12] assistant. >> So let's not just make.

[01:06:19] >> So let's not just make. Let's make something that

[01:06:20] Let's make something that matters.

[01:06:21] matters. ♪

[01:06:24] ♪ ♪

[01:06:40] ♪ [Applause]

[01:06:51] [Applause] >> SUNDAR PICHAI: Hello,

[01:06:56] >> SUNDAR PICHAI: Hello, Shoreline, and hello to everyone

[01:06:57] Shoreline, and hello to everyone watching from around the world.

[01:06:59] watching from around the world. Excited to be back for this

[01:07:01] Excited to be back for this year's I/O and what a year it's

[01:07:04] year's I/O and what a year it's been. All of the relentless

[01:07:08] been. All of the relentless shipping, the rapid advances in

[01:07:11] shipping, the rapid advances in technology, it's been a period

[01:07:15] technology, it's been a period of hyper-progress.

[01:07:15] of hyper-progress. I've definitely felt it myself.

[01:07:16] I've definitely felt it myself. It's been an intense year.

[01:07:17] It's been an intense year. Here's a look at what I've been

[01:07:21] Here's a look at what I've been up to.

[01:07:23] up to. [Laughter]

[01:07:24] [Laughter] Okay, I wish this was my year.

[01:07:29] Okay, I wish this was my year. Actually, the one of me plugging

[01:07:30] Actually, the one of me plugging in the TPUs is pretty accurate.

[01:07:34] in the TPUs is pretty accurate. But there's probably still more

[01:07:35] But there's probably still more work to do before it's in space,

[01:07:37] work to do before it's in space, we are working on that.

[01:07:41] we are working on that. On a serious note, it's been an

[01:07:42] On a serious note, it's been an extraordinary moment. It's been

[01:07:43] extraordinary moment. It's been 10 years since we pivoted the

[01:07:47] 10 years since we pivoted the company to be AI first. We knew

[01:07:50] company to be AI first. We knew how profound AI would be to

[01:07:51] how profound AI would be to advancing our mission, and

[01:07:51] advancing our mission, and improving people's lives at

[01:07:52] improving people's lives at scale.

[01:07:55] scale. This is why we are taking a

[01:07:58] This is why we are taking a differentiated, full-stack

[01:07:59] differentiated, full-stack approach to AI innovation.

[01:08:04] approach to AI innovation. From our custom silicon and

[01:08:05] From our custom silicon and secure foundation, to our

[01:08:06] secure foundation, to our world-class research and models,

[01:08:07] world-class research and models, to our products and platforms

[01:08:10] to our products and platforms that reach billions of people.

[01:08:10] that reach billions of people. This approach enables us to

[01:08:14] This approach enables us to iterate and innovate faster, and

[01:08:15] iterate and innovate faster, and it's lighting up every part of

[01:08:15] it's lighting up every part of the company.

[01:08:17] the company. What's really incredible is how

[01:08:19] What's really incredible is how people are using our AI.

[01:08:23] people are using our AI. Students prepping for final

[01:08:24] Students prepping for final exams through the Gemini app;

[01:08:25] exams through the Gemini app; musicians and artists using

[01:08:28] musicians and artists using generative AI modules like Lyria

[01:08:30] generative AI modules like Lyria and Veo as part of their

[01:08:31] and Veo as part of their creative flow; developers coding

[01:08:32] creative flow; developers coding and bringing their ideas to

[01:08:34] and bringing their ideas to life.

[01:08:34] life. I've been using Gemini in a

[01:08:36] I've been using Gemini in a myriad of ways.

[01:08:38] myriad of ways. Recently, I've been turning to

[01:08:41] Recently, I've been turning to Gemini to make sense of my

[01:08:42] Gemini to make sense of my parents' doctor visits. I'm

[01:08:43] parents' doctor visits. I'm sure many of you have done a

[01:08:45] sure many of you have done a version of that.

[01:08:47] version of that. These stories of how people are

[01:08:48] These stories of how people are using AI are the best measure of

[01:08:51] using AI are the best measure of progress.

[01:08:52] progress. To understand the scale at which

[01:08:54] To understand the scale at which people are adopting AI, there's

[01:08:55] people are adopting AI, there's another great proxy: Tokens,

[01:08:57] another great proxy: Tokens, the fundamental units of data

[01:08:59] the fundamental units of data our models process, many

[01:09:00] our models process, many representing a problem being

[01:09:03] representing a problem being solved.

[01:09:05] solved. Two years ago, we were

[01:09:06] Two years ago, we were processing 9.7 trillion tokens a

[01:09:14] processing 9.7 trillion tokens a month across our services, a

[01:09:16] month across our services, a huge number. Last year at I/O,

[01:09:17] huge number. Last year at I/O, that grew to about 480 trillion

[01:09:18] that grew to about 480 trillion tokens, impressive growth. Fast

[01:09:20] tokens, impressive growth. Fast forward to today, that number

[01:09:23] forward to today, that number has jumped seven times to

[01:09:23] has jumped seven times to 3.2 quadrillion tokens per

[01:09:24] 3.2 quadrillion tokens per month.

[01:09:25] month. [Applause]

[01:09:31] [Applause] Never imagined I would say

[01:09:35] Never imagined I would say quadrillion in an I/O keynote,

[01:09:36] quadrillion in an I/O keynote, but here we are.

[01:09:36] but here we are. Some out there might call this

[01:09:38] Some out there might call this token maxing, and there's

[01:09:39] token maxing, and there's probably some truth to it.

[01:09:41] probably some truth to it. I still think it tells an

[01:09:44] I still think it tells an important story, about our

[01:09:46] important story, about our products and how others are

[01:09:48] products and how others are building on it as well,

[01:09:49] building on it as well, especially our developers. Over

[01:09:51] especially our developers. Over 8.5 million of you are now

[01:09:51] 8.5 million of you are now building new apps and

[01:09:53] building new apps and experiences with our models

[01:09:56] experiences with our models monthly. And our model APIs are

[01:09:59] monthly. And our model APIs are now processing around 19 billion

[01:10:00] now processing around 19 billion tokens per minute.

[01:10:03] tokens per minute. Over the past 12 months, over

[01:10:06] Over the past 12 months, over 375 customers each processed

[01:10:10] 375 customers each processed more than 1 trillion tokens,

[01:10:11] more than 1 trillion tokens, representing incredible demand

[01:10:12] representing incredible demand for AI across our industry.

[01:10:18] for AI across our industry. We are, of course, also seeing

[01:10:19] We are, of course, also seeing incredible demand across our

[01:10:19] incredible demand across our products. We now have 13

[01:10:20] products. We now have 13 products with over 1 billion

[01:10:22] products with over 1 billion users each. Five of those have

[01:10:27] users each. Five of those have more than 3 billion users.

[01:10:28] more than 3 billion users. [Applause]

[01:10:31] [Applause] Our Gemini models are a big

[01:10:35] Our Gemini models are a big reason more people are using our

[01:10:37] reason more people are using our products, and why they are using

[01:10:39] products, and why they are using our products more. It all

[01:10:40] our products more. It all starts with Search, which is

[01:10:41] starts with Search, which is bringing the benefits of

[01:10:42] bringing the benefits of generative AI to more people

[01:10:43] generative AI to more people than any other product in the

[01:10:47] than any other product in the world.

[01:10:48] world. AI Overviews now has over

[01:10:48] AI Overviews now has over 2.5 billion monthly users, and

[01:10:51] 2.5 billion monthly users, and AI mode has been a revelation,

[01:10:54] AI mode has been a revelation, our biggest upgrade to Search

[01:10:55] our biggest upgrade to Search ever. People love it. In just

[01:10:56] ever. People love it. In just a year, it's already surpassed 1

[01:10:58] a year, it's already surpassed 1 billion monthly users.

[01:11:05] billion monthly users. When people use our AI-powered

[01:11:05] When people use our AI-powered features in Search, they use

[01:11:06] features in Search, they use Search more.

[01:11:06] Search more. I love how Search has become

[01:11:08] I love how Search has become less about individual queries

[01:11:11] less about individual queries and feels more like an ongoing

[01:11:12] and feels more like an ongoing conversation, giving you deeper

[01:11:13] conversation, giving you deeper insights and connecting you with

[01:11:16] insights and connecting you with the vastness of the Web.

[01:11:17] the vastness of the Web. Another place where we've been

[01:11:18] Another place where we've been rapidly innovating is in the

[01:11:22] rapidly innovating is in the Gemini app. Last year at I/O,

[01:11:23] Gemini app. Last year at I/O, the Gemini app had 400 million

[01:11:26] the Gemini app had 400 million monthly active users. Today, we

[01:11:27] monthly active users. Today, we have surpassed 900 million, more

[01:11:28] have surpassed 900 million, more than doubling in a year.

[01:11:30] than doubling in a year. [Applause]

[01:11:35] [Applause] In the same time, daily requests

[01:11:38] In the same time, daily requests have grown over seven times.

[01:11:40] have grown over seven times. It's incredible growth. We've

[01:11:42] It's incredible growth. We've been adding a lot of unique

[01:11:45] been adding a lot of unique features like Personal

[01:11:45] features like Personal Intelligence, which makes

[01:11:46] Intelligence, which makes responses more customized and

[01:11:47] responses more customized and helpful.

[01:11:48] helpful. And today, more than 50 billion

[01:11:50] And today, more than 50 billion images have been generated with

[01:11:51] images have been generated with our Nano Banana model.

[01:11:53] our Nano Banana model. It was a breakout star this past

[01:11:57] It was a breakout star this past year. I know you've all been

[01:11:58] year. I know you've all been having a lot of fun with it.

[01:11:58] having a lot of fun with it. Beyond the Gemini app, we are

[01:12:01] Beyond the Gemini app, we are also having much more natural

[01:12:02] also having much more natural conversations with Gemini,

[01:12:04] conversations with Gemini, directly inside many of our

[01:12:08] directly inside many of our products. Recently, Maps got

[01:12:10] products. Recently, Maps got its biggest upgrade in a decade,

[01:12:11] its biggest upgrade in a decade, including a new feature called

[01:12:14] including a new feature called Ask Maps. People are using it

[01:12:16] Ask Maps. People are using it to ask more complex and much

[01:12:20] to ask more complex and much longer questions.

[01:12:20] longer questions. Here's a real query from a

[01:12:21] Here's a real query from a parent: My kid just fell into

[01:12:24] parent: My kid just fell into the duckpond and the wedding

[01:12:27] the duckpond and the wedding starts in 30 minutes. Where can

[01:12:28] starts in 30 minutes. Where can I walk and buy her a new dress?

[01:12:32] I walk and buy her a new dress? I would like to hear how that

[01:12:35] I would like to hear how that turned out.

[01:12:37] turned out. We are also bringing this

[01:12:37] We are also bringing this conversational AI to two more

[01:12:40] conversational AI to two more products. First, Ask

[01:12:40] products. First, Ask YouTube. People come to

[01:12:41] YouTube. People come to YouTube every day to ask a

[01:12:43] YouTube every day to ask a lot of questions. There's a lot

[01:12:46] lot of questions. There's a lot of great videos, and sometimes,

[01:12:47] of great videos, and sometimes, it's hard to know where to

[01:12:48] it's hard to know where to start. Ask YouTube entirely

[01:12:52] start. Ask YouTube entirely reimagines the experience. Say

[01:12:53] reimagines the experience. Say you want to teach your

[01:12:55] you want to teach your 3-year-old how to ride a pedal

[01:12:59] 3-year-old how to ride a pedal bike, and they already know how

[01:13:03] bike, and they already know how to ride a balance bike. Just

[01:13:04] to ride a balance bike. Just ask YouTube. You will see a

[01:13:05] ask YouTube. You will see a couple of differences in

[01:13:05] couple of differences in results. The information is

[01:13:06] results. The information is digestible and easy to navigate.

[01:13:10] digestible and easy to navigate. You get an overview and helpful

[01:13:11] You get an overview and helpful tips. You will see videos that

[01:13:16] tips. You will see videos that best match your interests, so if

[01:13:17] best match your interests, so if you want to try a specific

[01:13:17] you want to try a specific method of teaching you can go

[01:13:23] method of teaching you can go deeper there. And best of all,

[01:13:23] deeper there. And best of all, it jumps right to the part of

[01:13:24] it jumps right to the part of the video most relevant for you,

[01:13:25] the video most relevant for you, brings back memories of teaching

[01:13:25] brings back memories of teaching my kids to ride.

[01:13:27] my kids to ride. And it remembers the context so

[01:13:29] And it remembers the context so you can follow up with questions

[01:13:35] you can follow up with questions like, should I buy one with

[01:13:36] like, should I buy one with handbrakes, or pedal brakes?

[01:13:36] handbrakes, or pedal brakes? Making it an ongoing

[01:13:37] Making it an ongoing conversation.

[01:13:38] conversation. It even lays out the information

[01:13:40] It even lays out the information in a table so it's easy to

[01:13:41] in a table so it's easy to compare. We are starting to

[01:13:44] compare. We are starting to test Ask YouTube now, and it

[01:13:45] test Ask YouTube now, and it will roll out broadly in the

[01:13:46] will roll out broadly in the U.S. this summer.

[01:13:51] U.S. this summer. [Applause]

[01:13:59] [Applause] So far, we have shown

[01:13:59] So far, we have shown conversational text queries.

[01:14:01] conversational text queries. There are a lot of times I want

[01:14:02] There are a lot of times I want to get things done at the speed

[01:14:03] to get things done at the speed of my voice. That is much more

[01:14:04] of my voice. That is much more possible today, thanks to

[01:14:05] possible today, thanks to technical leaps in our audio

[01:14:09] technical leaps in our audio model. A new feature called

[01:14:10] model. A new feature called Docs Live takes this to another

[01:14:13] Docs Live takes this to another level. To create a Doc with

[01:14:14] level. To create a Doc with Gemini before, you would have to

[01:14:16] Gemini before, you would have to type up a really precise prompt.

[01:14:19] type up a really precise prompt. With Docs Live, you can verbally

[01:14:22] With Docs Live, you can verbally brain-dump whatever is on your

[01:14:23] brain-dump whatever is on your mind and let Gemini do the rest.

[01:14:24] mind and let Gemini do the rest. Let's see it in action with the

[01:14:25] Let's see it in action with the demo from our product team.

[01:14:27] demo from our product team. This is all in real time, not

[01:14:28] This is all in real time, not sped up.

[01:14:32] sped up. >> All right.

[01:14:34] >> All right. Let's try this out.

[01:14:36] Let's try this out. So I just remembered I'm doing

[01:14:38] So I just remembered I'm doing an alumni talk for my high

[01:14:40] an alumni talk for my high school's career day tomorrow,

[01:14:41] school's career day tomorrow, and I need to come up with some

[01:14:41] and I need to come up with some talking points to explain what

[01:14:42] talking points to explain what do for a living as a software

[01:14:43] do for a living as a software engineer, but I'm not really

[01:14:44] engineer, but I'm not really sure where to start.

[01:14:52] sure where to start. Actually, can you pull my resume

[01:14:53] Actually, can you pull my resume from Drive?

[01:14:54] from Drive? Although that might be boring.

[01:14:54] Although that might be boring. Maybe can you come up with some

[01:14:57] Maybe can you come up with some funny analogies so it will be a

[01:14:58] funny analogies so it will be a more engaging talk for the

[01:15:02] more engaging talk for the students? And also, I think the

[01:15:02] students? And also, I think the school sent me an e-mail. I

[01:15:03] school sent me an e-mail. I think the subject is something

[01:15:06] think the subject is something like "career day logistics."

[01:15:09] like "career day logistics." Maybe grab the details from

[01:15:09] Maybe grab the details from there, throw them at the top of

[01:15:12] there, throw them at the top of the Doc, so I know where to go

[01:15:12] the Doc, so I know where to go and what time to get there.

[01:15:13] and what time to get there. Let's update those requirements

[01:15:14] Let's update those requirements and turn this straight into a

[01:15:15] and turn this straight into a draft.

[01:15:23] draft. [Applause]

[01:15:26] [Applause] >> This is cool, but it's a

[01:15:29] >> This is cool, but it's a little dense.

[01:15:29] little dense. Maybe format the analogies as a

[01:15:30] Maybe format the analogies as a table, so it's a little easier

[01:15:31] table, so it's a little easier for me to scan and also add a

[01:15:31] for me to scan and also add a note to tell the story about ho

[01:15:32] note to tell the story about ho my brother inspired me to becom

[01:15:33] my brother inspired me to becom a software engineer.

[01:15:37] a software engineer. Sort of at the top of my Doc an

[01:15:38] Sort of at the top of my Doc an bold it so I don't miss it.

[01:15:45] bold it so I don't miss it. [Applause]

[01:15:45] [Applause] Yeah, that looks great.

[01:15:54] Yeah, that looks great. >> SUNDAR PICHAI: In the future,

[01:15:56] >> SUNDAR PICHAI: In the future, you'll be able to create new

[01:15:57] you'll be able to create new Docs and edit them directly all

[01:16:00] Docs and edit them directly all with your voice.

[01:16:01] with your voice. Docs Live is rolling out for Pro

[01:16:02] Docs Live is rolling out for Pro and Ultra subscribers this

[01:16:05] and Ultra subscribers this summer and the same powerful

[01:16:07] summer and the same powerful voice capabilities will come to

[01:16:08] voice capabilities will come to Gmail and Google Keep then, too.

[01:16:12] Gmail and Google Keep then, too. [Applause]

[01:16:15] [Applause] Incredible to see the pace of

[01:16:17] Incredible to see the pace of innovation rolling out across

[01:16:20] innovation rolling out across our products.

[01:16:21] our products. Supporting all of this at scale

[01:16:23] Supporting all of this at scale for our users, while also

[01:16:24] for our users, while also serving enterprises and

[01:16:25] serving enterprises and developers around the world

[01:16:27] developers around the world requires massive investments in

[01:16:29] requires massive investments in infrastructure, and we've been

[01:16:33] infrastructure, and we've been investing for today and for the

[01:16:34] investing for today and for the future.

[01:16:34] future. In 2022, we were spending $31

[01:16:35] In 2022, we were spending $31 billion annually in capex.

[01:16:38] billion annually in capex. This year, we expect that number

[01:16:43] This year, we expect that number to be about six times that,

[01:16:47] to be about six times that, approximately $180 to $190

[01:16:48] approximately $180 to $190 billion. A key part of this

[01:16:49] billion. A key part of this investment is our custom

[01:16:50] investment is our custom silicon. A decade ago, we

[01:16:52] silicon. A decade ago, we announced our very first

[01:16:53] announced our very first commercial tensor processing

[01:16:57] commercial tensor processing unit or TPU on this I/O stage.

[01:16:58] unit or TPU on this I/O stage. Since then, we have transformed

[01:17:01] Since then, we have transformed how the industry builds for AI.

[01:17:03] how the industry builds for AI. We recently announced our eighth

[01:17:04] We recently announced our eighth generation of TPUs at Cloud

[01:17:05] generation of TPUs at Cloud Next.

[01:17:06] Next. For the first time, we have

[01:17:11] For the first time, we have taken a dual-chip approach with

[01:17:11] taken a dual-chip approach with specialized architectures for

[01:17:14] specialized architectures for training and inference: TPU 8t

[01:17:15] training and inference: TPU 8t and 8i.

[01:17:16] and 8i. While they may look similar,

[01:17:17] While they may look similar, they're actually pretty

[01:17:19] they're actually pretty different. 8t is optimized for

[01:17:24] different. 8t is optimized for large-scale pretraining, and

[01:17:24] large-scale pretraining, and it's nearly three times the raw

[01:17:25] it's nearly three times the raw computing power of our previous

[01:17:27] computing power of our previous generation.

[01:17:28] generation. And we've taken a fundamentally

[01:17:29] And we've taken a fundamentally different approach with our

[01:17:30] different approach with our training infrastructure. With

[01:17:31] training infrastructure. With JAX and Pathways, our training

[01:17:33] JAX and Pathways, our training is no longer constrained by the

[01:17:36] is no longer constrained by the limits of a single massive data

[01:17:39] limits of a single massive data center. Instead, we can now

[01:17:39] center. Instead, we can now seamlessly distribute training

[01:17:41] seamlessly distribute training across multiple sites, scaling

[01:17:44] across multiple sites, scaling across more than 1 million TPUs

[01:17:47] across more than 1 million TPUs globally.

[01:17:48] globally. This gives us the ability to

[01:17:49] This gives us the ability to create the largest training

[01:17:50] create the largest training cluster in the world. For model

[01:17:52] cluster in the world. For model builders, this means training

[01:17:53] builders, this means training larger, more capable models in

[01:17:55] larger, more capable models in weeks rather than months.

[01:17:59] weeks rather than months. TPU 8i is designed for

[01:18:02] TPU 8i is designed for inference. We have dramatically

[01:18:04] inference. We have dramatically improved speed at every step,

[01:18:05] improved speed at every step, because if you learn anything in

[01:18:08] because if you learn anything in 27 years of working on Search,

[01:18:11] 27 years of working on Search, it's that latency matters. To

[01:18:12] it's that latency matters. To give you a live sense of what

[01:18:13] give you a live sense of what the speed feels like, here's a

[01:18:18] the speed feels like, here's a Prompt on an upcoming Flash

[01:18:21] Prompt on an upcoming Flash model if it were running on 8i.

[01:18:24] model if it were running on 8i. I'll is to create a Chrome Dino

[01:18:28] I'll is to create a Chrome Dino game, Push Summit. The response

[01:18:30] game, Push Summit. The response is generated in real time.

[01:18:33] is generated in real time. As you watch, take a look at the

[01:18:34] As you watch, take a look at the tokens per second in the top

[01:18:34] tokens per second in the top right corner.

[01:18:35] right corner. The speed is pretty incredible,

[01:18:37] The speed is pretty incredible, nearly 1,500 tokens per second.

[01:18:39] nearly 1,500 tokens per second. It almost took longer to write

[01:18:40] It almost took longer to write out the request, and the game is

[01:18:42] out the request, and the game is pretty fun, too.

[01:18:47] pretty fun, too. [Applause]

[01:18:50] [Applause] In addition to speed, we're also

[01:18:52] In addition to speed, we're also thinking about scaling

[01:18:53] thinking about scaling sustainably. Both chips are

[01:18:54] sustainably. Both chips are more energy efficient,

[01:18:57] more energy efficient, delivering up to two times

[01:18:58] delivering up to two times better performance per watt.

[01:19:01] better performance per watt. TPUs have been hard at work

[01:19:03] TPUs have been hard at work training for I/O this year. I'm

[01:19:05] training for I/O this year. I'm told we have a behind-the-scenes

[01:19:06] told we have a behind-the-scenes look.

[01:19:19] look. >> Hey, good weekend?

[01:19:19] >> Hey, good weekend? >> Yeah.

[01:19:20] >> Yeah. Just folded proteins across rar

[01:19:22] Just folded proteins across rar oncology data sets.

[01:19:22] oncology data sets. You?

[01:19:24] You? >> Simulated the next 50 years

[01:19:25] >> Simulated the next 50 years of climate data.

[01:19:25] of climate data. >> I made a picture of a pug.

[01:19:26] >> I made a picture of a pug. >> What?

[01:19:29] >> What? >> You ever a see pug dressed

[01:19:30] >> You ever a see pug dressed like an accountant?

[01:19:31] like an accountant? >> No.

[01:19:36] >> No. >> You wanna?

[01:19:37] >> You wanna? >> All right, listen up TPUs.

[01:19:38] >> All right, listen up TPUs. I/O is starting soon, and we go

[01:19:40] I/O is starting soon, and we go a job to do.

[01:19:40] a job to do. Actually, we got trillions of

[01:19:41] Actually, we got trillions of jobs to do.

[01:19:43] jobs to do. So clear cache.

[01:19:44] So clear cache. Timmy!

[01:19:46] Timmy! >> Huh?

[01:19:46] >> Huh? >> Dry your fans and let's fire

[01:19:46] >> Dry your fans and let's fire it up.

[01:19:48] it up. ♪

[01:19:48] ♪ ♪

[01:20:11] ♪ >> Hey, what are you doing?

[01:20:12] >> Hey, what are you doing? >> I'm doing the montage thing.

[01:20:14] >> I'm doing the montage thing. >> Yeah, well, how about you

[01:20:15] >> Yeah, well, how about you montage your way down here and

[01:20:15] montage your way down here and help out?

[01:20:16] help out? >> What?

[01:20:19] >> What? Like now?

[01:20:19] Like now? All right, coming.

[01:20:21] All right, coming. ♪

[01:20:23] ♪ ♪

[01:20:37] ♪ >> SUNDAR PICHAI: I bet Timmy

[01:20:38] >> SUNDAR PICHAI: I bet Timmy TPU will be ready to teraflop

[01:20:41] TPU will be ready to teraflop right into bed after I/O.

[01:20:43] right into bed after I/O. Our compute innovations enable

[01:20:46] Our compute innovations enable our advancements. There are

[01:20:47] our advancements. There are three areas where I want to go

[01:20:48] three areas where I want to go deeper today to show you the

[01:20:51] deeper today to show you the progress in each: Models,

[01:20:51] progress in each: Models, coding, and agents.

[01:20:56] coding, and agents. Let's start with the exciting

[01:20:58] Let's start with the exciting progress in world models. With

[01:20:59] progress in world models. With world models, AI is moving from

[01:21:01] world models, AI is moving from predicting text to simulating

[01:21:03] predicting text to simulating reality. Demis and the team at

[01:21:05] reality. Demis and the team at Google DeepMind have been

[01:21:06] Google DeepMind have been working to push the boundaries

[01:21:08] working to push the boundaries of what these models can do.

[01:21:11] of what these models can do. Let me invite Demis out to share

[01:21:11] Let me invite Demis out to share more.

[01:21:13] more. ♪

[01:21:13] ♪ ♪

[01:21:14] ♪ [Applause]

[01:21:25] [Applause] >> DEMIS HASSABIS: Hi, everyone

[01:21:26] >> DEMIS HASSABIS: Hi, everyone It's really great to be here.

[01:21:29] It's really great to be here. Over the past year, AI

[01:21:30] Over the past year, AI capabilities have leaped

[01:21:32] capabilities have leaped forwards. We now have agents

[01:21:33] forwards. We now have agents that can plan and act on our

[01:21:36] that can plan and act on our behalf. And artificial general

[01:21:37] behalf. And artificial general intelligence is just a few years

[01:21:39] intelligence is just a few years away.

[01:21:40] away. Today, I'm excited to share the

[01:21:43] Today, I'm excited to share the progress we've made toward

[01:21:46] progress we've made toward building AGI. Last year, I

[01:21:48] building AGI. Last year, I outlined our vision of extending

[01:21:49] outlined our vision of extending Gemini's incredible multimodal

[01:21:50] Gemini's incredible multimodal capabilities to become a world

[01:21:54] capabilities to become a world model. AI that can understand

[01:21:57] model. AI that can understand and simulate the world. This is

[01:21:59] and simulate the world. This is a crucial aspect of achieving

[01:22:00] a crucial aspect of achieving AGI and will be important for

[01:22:03] AGI and will be important for everything from building AI

[01:22:03] everything from building AI assistants to training robots.

[01:22:06] assistants to training robots. Now, we're taking the next big

[01:22:08] Now, we're taking the next big step.

[01:22:10] step. I'm excited to announce Gemini

[01:22:17] I'm excited to announce Gemini Omni.

[01:22:17] Omni. [Applause]

[01:22:22] [Applause] Our new model that can create

[01:22:23] Our new model that can create anything from any input. It

[01:22:23] anything from any input. It combines Gemini's intelligence

[01:22:25] combines Gemini's intelligence with the best of our generative

[01:22:26] with the best of our generative media models for a new level of

[01:22:27] media models for a new level of world understanding,

[01:22:30] world understanding, multimodality and editing.

[01:22:32] multimodality and editing. Models like Veo, Nano Banana and

[01:22:35] Models like Veo, Nano Banana and Genie are able to create

[01:22:36] Genie are able to create extremely realistic videos,

[01:22:39] extremely realistic videos, images and interactive

[01:22:39] images and interactive simulations. Although not

[01:22:41] simulations. Although not perfect, they already

[01:22:42] perfect, they already demonstrate some impressive

[01:22:44] demonstrate some impressive notions of intuitive physics.

[01:22:46] notions of intuitive physics. And with Omni, we've now made

[01:22:49] And with Omni, we've now made even more progress. It's a

[01:22:50] even more progress. It's a step-change in simulating things

[01:22:53] step-change in simulating things like kinetic energy and gravity.

[01:22:54] like kinetic energy and gravity. Previous systems would have

[01:22:55] Previous systems would have found these concepts difficult.

[01:22:59] found these concepts difficult. Gemini's world knowledge and

[01:23:00] Gemini's world knowledge and reasoning really shine in Omni.

[01:23:03] reasoning really shine in Omni. It can translate complex ideas

[01:23:05] It can translate complex ideas into highly accurate videos.

[01:23:07] into highly accurate videos. So, for example, you can give it

[01:23:11] So, for example, you can give it a simple prompt like "make a

[01:23:12] a simple prompt like "make a claymation explainer of protein

[01:23:14] claymation explainer of protein folding" and get this.

[01:23:14] folding" and get this. >> Proteins start as chains of

[01:23:15] >> Proteins start as chains of amino acids. They fold into

[01:23:17] amino acids. They fold into patterns, like the alpha helix

[01:23:19] patterns, like the alpha helix and flat sections called beta

[01:23:25] and flat sections called beta sheets forming a perfect

[01:23:26] sheets forming a perfect three-dimensional shape.

[01:23:26] three-dimensional shape. >> DEMIS HASSABIS: But the

[01:23:27] >> DEMIS HASSABIS: But the initial generation is just the

[01:23:29] initial generation is just the start. The creative process is

[01:23:31] start. The creative process is rarely a single step; it's

[01:23:31] rarely a single step; it's usually iterative. Just like

[01:23:33] usually iterative. Just like Nano Banana redefined image

[01:23:35] Nano Banana redefined image editing, Omni gives you a much

[01:23:35] editing, Omni gives you a much more natural way to edit video

[01:23:38] more natural way to edit video with conversational language.

[01:23:42] with conversational language. What's really cool is you can

[01:23:46] What's really cool is you can give it your own videos -- for

[01:23:49] give it your own videos -- for example, this selfie -- and

[01:23:50] example, this selfie -- and change reality in a really fun

[01:23:51] change reality in a really fun way. You can easily adjust the

[01:23:54] way. You can easily adjust the details and style, or even add

[01:23:59] details and style, or even add elements. And the whole scene

[01:24:00] elements. And the whole scene morphs to reflect your new idea.

[01:24:03] morphs to reflect your new idea. [Applause]

[01:24:07] [Applause] A simple circle turns into a

[01:24:08] A simple circle turns into a black hole or an evening stroll

[01:24:10] black hole or an evening stroll comes to life. Anything becomes

[01:24:11] comes to life. Anything becomes a canvas for creating entirely

[01:24:13] a canvas for creating entirely new realities.

[01:24:15] new realities. Let's take a look at what Omni

[01:24:15] Let's take a look at what Omni can do.

[01:24:17] can do. ♪

[01:24:18] ♪ ♪

[01:25:04] ♪ >> DEMIS HASSABIS: We're

[01:25:05] >> DEMIS HASSABIS: We're starting with video, but over

[01:25:05] starting with video, but over time, Omni will be able to

[01:25:06] time, Omni will be able to generate any output from any

[01:25:07] generate any output from any input. This was always our goal

[01:25:08] input. This was always our goal with Gemini, and why we built it

[01:25:11] with Gemini, and why we built it to be multimodal from the very

[01:25:15] to be multimodal from the very start.

[01:25:25] start. It was a harder path but the

[01:25:26] It was a harder path but the foundation is now paying off.

[01:25:27] foundation is now paying off. Today, we're launching the first

[01:25:28] Today, we're launching the first model in the Omni family, Gemini

[01:25:29] model in the Omni family, Gemini Omni Flash. It's now available

[01:25:30] Omni Flash. It's now available across our products, and you'll

[01:25:31] across our products, and you'll hear more about this later.

[01:25:31] hear more about this later. We're excited about the progress

[01:25:32] We're excited about the progress we're making and we'll be

[01:25:33] we're making and we'll be sharing more about Omni Pro

[01:25:34] sharing more about Omni Pro soon.

[01:25:34] soon. We can't wait to see what you

[01:25:35] We can't wait to see what you create.

[01:25:35] create. Back to you, Sundar.

[01:25:36] Back to you, Sundar. [Applause]

[01:25:45] [Applause] >> SUNDAR PICHAI: Thanks, Demis.

[01:25:49] >> SUNDAR PICHAI: Thanks, Demis. There's huge progress.

[01:25:50] There's huge progress. As generative AI gets better, so

[01:25:52] As generative AI gets better, so does the need for greater

[01:25:54] does the need for greater transparency. Research shows

[01:25:55] transparency. Research shows people can correctly identify

[01:25:56] people can correctly identify high-quality deepfake videos

[01:26:00] high-quality deepfake videos only about a quarter of the

[01:26:01] only about a quarter of the time. Three years ago, we

[01:26:03] time. Three years ago, we launched SynthID, our watermark

[01:26:05] launched SynthID, our watermark that is invisible to the naked

[01:26:06] that is invisible to the naked eye. Since launch, SynthID has

[01:26:10] eye. Since launch, SynthID has now watermarked 100 billion

[01:26:13] now watermarked 100 billion images and videos, along with

[01:26:13] images and videos, along with 60,000 years of audio assets.

[01:26:16] 60,000 years of audio assets. Millions of people are using our

[01:26:21] Millions of people are using our SynthID Detector in the Gemini

[01:26:22] SynthID Detector in the Gemini app to verify AI-generated

[01:26:23] app to verify AI-generated content. And now, we are going

[01:26:26] content. And now, we are going a step further and adding

[01:26:29] a step further and adding Content Credentials Verification

[01:26:30] Content Credentials Verification across products. This will show

[01:26:32] across products. This will show you if the origin of the content

[01:26:33] you if the origin of the content was AI or a camera, and if it

[01:26:35] was AI or a camera, and if it has been edited with generative

[01:26:36] has been edited with generative AI tools.

[01:26:40] AI tools. In this example, Gemini can tell

[01:26:41] In this example, Gemini can tell this photo was captured with a

[01:26:47] this photo was captured with a Pixel camera and then edited

[01:26:48] Pixel camera and then edited with Google Photos.

[01:26:49] with Google Photos. We want more people to have easy

[01:26:50] We want more people to have easy access to these tools.

[01:26:52] access to these tools. So we are expanding both SynthID

[01:26:54] So we are expanding both SynthID and Content Credentials

[01:26:55] and Content Credentials Verification to Search and

[01:26:59] Verification to Search and Chrome.

[01:27:00] Chrome. [Applause]

[01:27:05] [Applause] You can simply circle to search

[01:27:07] You can simply circle to search or right-click in Chrome and ask

[01:27:10] or right-click in Chrome and ask "was this generated with AI?"

[01:27:11] "was this generated with AI?" And you'll get a clear response

[01:27:12] And you'll get a clear response along with other helpful

[01:27:13] along with other helpful context.

[01:27:13] context. For example, this image was

[01:27:14] For example, this image was making the rounds on social

[01:27:18] making the rounds on social media last year.

[01:27:20] media last year. [Laughter]

[01:27:22] [Laughter] It's obviously fake. I don't

[01:27:24] It's obviously fake. I don't eat hamburgers.

[01:27:28] eat hamburgers. It might not be as clear to

[01:27:28] It might not be as clear to everyone else. That's where

[01:27:30] everyone else. That's where these tools can be really

[01:27:33] these tools can be really useful. Of course, this only

[01:27:33] useful. Of course, this only works at scale if more partners

[01:27:34] works at scale if more partners decide to watermark their own

[01:27:35] decide to watermark their own AI-generated content.

[01:27:38] AI-generated content. NVIDIA signed on to SynthID last

[01:27:41] NVIDIA signed on to SynthID last year. And today, I'm thrilled

[01:27:42] year. And today, I'm thrilled to announce that Open AI, Kakao

[01:27:46] to announce that Open AI, Kakao and Eleven Labs are adopting

[01:27:47] and Eleven Labs are adopting SynthID, too.

[01:27:47] SynthID, too. [Applause]

[01:27:53] [Applause] It's great to see the

[01:27:54] It's great to see the cross-industry collaboration.

[01:27:55] cross-industry collaboration. And we're looking forward to

[01:27:57] And we're looking forward to expanding to more partners and

[01:27:59] expanding to more partners and setting the standard for

[01:28:05] setting the standard for transparency for the AI era.

[01:28:06] transparency for the AI era. That's a look at the progress we

[01:28:08] That's a look at the progress we are making with World Models.

[01:28:09] are making with World Models. Now, let's talk about what's

[01:28:11] Now, let's talk about what's next for our Gemini 3 Family.

[01:28:12] next for our Gemini 3 Family. Gemini 3 launched a few months

[01:28:13] Gemini 3 launched a few months ago with a full family of

[01:28:15] ago with a full family of models. It's been our most

[01:28:16] models. It's been our most adopted series yet. We have

[01:28:19] adopted series yet. We have loved seeing developers use

[01:28:21] loved seeing developers use Flash as a daily driver and

[01:28:23] Flash as a daily driver and build incredible experiences

[01:28:25] build incredible experiences with Pro's deep reasoning and

[01:28:26] with Pro's deep reasoning and multimodal capabilities. We've

[01:28:28] multimodal capabilities. We've been hard at work on improving

[01:28:31] been hard at work on improving these models, especially focused

[01:28:37] these models, especially focused on agentic coding, long-horizon

[01:28:38] on agentic coding, long-horizon tasks and real-world workflows.

[01:28:38] tasks and real-world workflows. And today, I'm excited to

[01:28:43] And today, I'm excited to introduce Gemini 3.5 Flash, our

[01:28:43] introduce Gemini 3.5 Flash, our first in a series of models

[01:28:48] first in a series of models combining frontier intelligence

[01:28:50] combining frontier intelligence with action.

[01:28:51] with action. Two things I would highlight.

[01:28:55] Two things I would highlight. First, when compared to 3.1 Pr

[01:29:00] First, when compared to 3.1 Pr Flash is better across the board

[01:29:01] Flash is better across the board in almost all benchmarks.

[01:29:04] in almost all benchmarks. It's made huge progress in

[01:29:06] It's made huge progress in coding, and look at that

[01:29:12] coding, and look at that extraordinary jump in GDP val

[01:29:16] extraordinary jump in GDP val a benchmark that captures many

[01:29:19] a benchmark that captures many real-world economically valuable

[01:29:19] real-world economically valuable tasks.

[01:29:19] tasks. Second, 3.5 Flash is a very

[01:29:20] Second, 3.5 Flash is a very capable model, at the frontier

[01:29:21] capable model, at the frontier and comparable to the best

[01:29:26] and comparable to the best models, but much faster, which

[01:29:27] models, but much faster, which is why when you look at the

[01:29:27] is why when you look at the intelligence versus output

[01:29:28] intelligence versus output speed, it's in a league of its

[01:29:29] speed, it's in a league of its own in the top right quadrant.

[01:29:33] own in the top right quadrant. When looking at output tokens

[01:29:36] When looking at output tokens per second, it's four times

[01:29:36] per second, it's four times faster than other frontier

[01:29:37] faster than other frontier models, and it's an incredible

[01:29:38] models, and it's an incredible delight to use.

[01:29:39] delight to use. The new model has been a

[01:29:40] The new model has been a game-changer for us internally

[01:29:40] game-changer for us internally at Google. We've been using 3.5

[01:29:42] at Google. We've been using 3.5 Flash with the reimagined

[01:29:45] Flash with the reimagined version of our agent-first

[01:29:47] version of our agent-first development platform,

[01:29:49] development platform, Antigravity. And it's

[01:29:52] Antigravity. And it's dramatically accelerated how we

[01:29:53] dramatically accelerated how we build.

[01:29:54] build. In March, we were processing

[01:29:55] In March, we were processing half a trillion tokens a day

[01:29:59] half a trillion tokens a day internally for our developers.

[01:29:59] internally for our developers. We've been doubling every few

[01:30:01] We've been doubling every few weeks. And now, we are

[01:30:02] weeks. And now, we are processing more than 3 trillion

[01:30:03] processing more than 3 trillion tokens a day.

[01:30:05] tokens a day. That scale has created a

[01:30:07] That scale has created a powerful feedback loop, which is

[01:30:11] powerful feedback loop, which is helping us improve 3.5.

[01:30:12] helping us improve 3.5. And, of course, we are bringing

[01:30:14] And, of course, we are bringing it today to developers in

[01:30:14] it today to developers in Antigravity. Varun is going to

[01:30:15] Antigravity. Varun is going to share more.

[01:30:17] share more. ♪

[01:30:17] ♪ ♪

[01:30:18] ♪ [Applause]

[01:30:25] [Applause] >> VARUN MOHAN: It's truly an

[01:30:26] >> VARUN MOHAN: It's truly an amazing time to be a builder.

[01:30:27] amazing time to be a builder. We've moved beyond AI tools that

[01:30:30] We've moved beyond AI tools that help us write, to agents that

[01:30:31] help us write, to agents that help us act.

[01:30:32] help us act. These agents have lowered the

[01:30:35] These agents have lowered the barrier to development so much

[01:30:35] barrier to development so much that anyone can be a builder,

[01:30:38] that anyone can be a builder, even busy CEOs. In fact, Sundar

[01:30:41] even busy CEOs. In fact, Sundar used Google Antigravity last

[01:30:42] used Google Antigravity last week to fix a bug in the Google

[01:30:47] week to fix a bug in the Google codebase. When we launched the

[01:30:50] codebase. When we launched the Antigravity IDE in November, we

[01:30:51] Antigravity IDE in November, we made sure to nail the core

[01:30:52] made sure to nail the core agent-powered IDE experience and

[01:30:57] agent-powered IDE experience and added an experimental

[01:30:59] added an experimental first-of-its-kind agent for

[01:31:00] first-of-its-kind agent for Surface as a glimpse of where we

[01:31:01] Surface as a glimpse of where we are heading.

[01:31:01] are heading. Millions of you already use

[01:31:05] Millions of you already use Antigravity, and so we're

[01:31:06] Antigravity, and so we're Excited to bring you even more

[01:31:09] Excited to bring you even more today.

[01:31:09] today. We've seen the diversity of

[01:31:10] We've seen the diversity of tasks, preferences and, frankly,

[01:31:11] tasks, preferences and, frankly, product feedback, and we've

[01:31:11] product feedback, and we've taken all of these learnings,

[01:31:13] taken all of these learnings, and now, Antigravity is

[01:31:17] and now, Antigravity is massively expanding its suite of

[01:31:18] massively expanding its suite of agentic capabilities, surfaces,

[01:31:21] agentic capabilities, surfaces, integration, and product

[01:31:21] integration, and product features.

[01:31:24] features. To start, we're launching a full

[01:31:33] To start, we're launching a full CLI experience: an Antigravity

[01:31:34] CLI experience: an Antigravity SDK, native voice support with

[01:31:35] SDK, native voice support with Gemini audio models and

[01:31:35] Gemini audio models and integration with many surfaces

[01:31:36] integration with many surfaces and platforms, like Android,

[01:31:39] and platforms, like Android, Firebase, and Google AI Studio.

[01:31:40] Firebase, and Google AI Studio. [Applause]

[01:31:48] [Applause] All of this is available for you

[01:31:49] All of this is available for you to try today. Most importantly,

[01:31:52] to try today. Most importantly, at the core is Antigravity 2.0,

[01:31:55] at the core is Antigravity 2.0, a new stand-alone desktop

[01:31:56] a new stand-alone desktop application that delivers fully

[01:31:57] application that delivers fully on that original glimpse of a

[01:32:00] on that original glimpse of a truly agent-optimized

[01:32:03] truly agent-optimized experience.

[01:32:04] experience. The new Antigravity is

[01:32:05] The new Antigravity is unabashedly agent first,

[01:32:07] unabashedly agent first, focusing on the core agent

[01:32:07] focusing on the core agent conversation, agent-produced

[01:32:10] conversation, agent-produced artifacts and multi-agent

[01:32:15] artifacts and multi-agent orchestration.

[01:32:16] orchestration. Like I said: Unabashedly agent

[01:32:17] Like I said: Unabashedly agent first.

[01:32:17] first. As Sundar mentioned, this is the

[01:32:18] As Sundar mentioned, this is the exact experience teams here at

[01:32:19] exact experience teams here at Google have been using to drive

[01:32:21] Google have been using to drive massive value.

[01:32:24] massive value. The Antigravity Agent Harness,

[01:32:24] The Antigravity Agent Harness, the invisible framework for

[01:32:25] the invisible framework for Gemini to perform real-world

[01:32:27] Gemini to perform real-world tasks, has become much more

[01:32:30] tasks, has become much more powerful with new core

[01:32:31] powerful with new core primitives, such as subagents,

[01:32:32] primitives, such as subagents, hooks, and asynchronous task

[01:32:36] hooks, and asynchronous task management.

[01:32:36] management. And underpinning all of this are

[01:32:38] And underpinning all of this are the Gemini models, with Gemini

[01:32:40] the Gemini models, with Gemini 3.5 Flash having been

[01:32:42] 3.5 Flash having been co-optimized with the

[01:32:44] co-optimized with the Antigravity Harness.

[01:32:47] Antigravity Harness. Of course, being engineers, we

[01:32:48] Of course, being engineers, we were curious to see how far we

[01:32:49] were curious to see how far we could push the limits of what

[01:32:49] could push the limits of what was possible with these agents

[01:32:51] was possible with these agents and models.

[01:32:52] and models. So using the new Antigravity and

[01:32:55] So using the new Antigravity and Gemini 3.5 Flash, we asked our

[01:32:57] Gemini 3.5 Flash, we asked our agents to take on what we

[01:32:59] agents to take on what we consider to be a highly complex

[01:33:03] consider to be a highly complex and impressive task: Build a

[01:33:06] and impressive task: Build a working operating system from

[01:33:07] working operating system from scratch. We were surprised by

[01:33:07] scratch. We were surprised by what we found.

[01:33:10] what we found. Asynchronously, Antigravity

[01:33:12] Asynchronously, Antigravity broke down the challenge into a

[01:33:13] broke down the challenge into a cohesive plan, tackled tasks via

[01:33:14] cohesive plan, tackled tasks via parallel subagents, generated,

[01:33:16] parallel subagents, generated, executed, and iterated over its

[01:33:20] executed, and iterated over its very own tests. Over 12 hours,

[01:33:23] very own tests. Over 12 hours, 93 subagents working in parallel

[01:33:27] 93 subagents working in parallel made over 15,000 model requests

[01:33:29] made over 15,000 model requests and processed 2.6 billion tokens

[01:33:31] and processed 2.6 billion tokens to take an initially empty

[01:33:32] to take an initially empty project to the core of a

[01:33:33] project to the core of a functioning operating system

[01:33:38] functioning operating system [Applause]

[01:33:45] [Applause] This was not possible on Gemini

[01:33:50] This was not possible on Gemini 3.1 Pro, but thanks to the

[01:33:50] 3.1 Pro, but thanks to the performance and cost

[01:33:51] performance and cost efficiencies of Gemini 3.5

[01:33:52] efficiencies of Gemini 3.5 Flash, building an entirely

[01:33:53] Flash, building an entirely functional operating system

[01:33:55] functional operating system consumed less than $1,000 of API

[01:33:59] consumed less than $1,000 of API credits.

[01:34:05] credits. [Applause]

[01:34:07] [Applause] The Antigravity agents wrote

[01:34:08] The Antigravity agents wrote every line of code from the

[01:34:09] every line of code from the scheduler to the memory

[01:34:15] scheduler to the memory management to the file system;

[01:34:15] management to the file system; generated, audited and tested

[01:34:16] generated, audited and tested entirely by an autonomous team

[01:34:17] entirely by an autonomous team of agents.

[01:34:18] of agents. To put this in context,

[01:34:23] To put this in context, developing an OS from scratch is

[01:34:29] developing an OS from scratch is notoriously brutal and can take

[01:34:29] notoriously brutal and can take many months to build.

[01:34:30] many months to build. We weren't just building an

[01:34:31] We weren't just building an application, but a functioning

[01:34:31] application, but a functioning operating system that

[01:34:32] operating system that applications can run on.

[01:34:33] applications can run on. Let's take this live and

[01:34:34] Let's take this live and actually show the operating

[01:34:35] actually show the operating system in action.

[01:34:38] system in action. So here, I'm actually in a

[01:34:39] So here, I'm actually in a terminal window in the OS that

[01:34:41] terminal window in the OS that Antigravity built. It's not

[01:34:43] Antigravity built. It's not super easy to demo a working

[01:34:44] super easy to demo a working operating system so let's try

[01:34:45] operating system so let's try something fun to see if it

[01:34:47] something fun to see if it works.

[01:34:47] works. One of the interesting utilities

[01:34:49] One of the interesting utilities you can install is SL, a common

[01:34:52] you can install is SL, a common typo for LS. Without spoiling

[01:34:53] typo for LS. Without spoiling it, here it goes.

[01:34:58] it, here it goes. It works. You could see a cool

[01:34:59] It works. You could see a cool locomotive passing through the

[01:35:01] locomotive passing through the screen with the Antigravity logo

[01:35:04] screen with the Antigravity logo on it. But clearly, this isn't

[01:35:07] on it. But clearly, this isn't a real OS unless I can play

[01:35:07] a real OS unless I can play Doom.

[01:35:13] Doom. So I try running Doom right now,

[01:35:16] So I try running Doom right now, it just doesn't work. Turns out

[01:35:21] it just doesn't work. Turns out that this is missing some

[01:35:21] that this is missing some necessary video and keyboard

[01:35:23] necessary video and keyboard drivers.

[01:35:26] drivers. So let's just try to fix it in

[01:35:28] So let's just try to fix it in the new Antigravity.

[01:35:29] the new Antigravity. I have a prompt prepared and I'm

[01:35:31] I have a prompt prepared and I'm going to paste it in.

[01:35:36] going to paste it in. While it's running, let's take a

[01:35:40] While it's running, let's take a tour in Antigravity 2.0.

[01:35:41] tour in Antigravity 2.0. As you can see, Antigravity 2.0

[01:35:41] As you can see, Antigravity 2.0 is fully agent first with all

[01:35:42] is fully agent first with all the agent conversations on the

[01:35:43] the agent conversations on the side and all the projects, as

[01:35:44] side and all the projects, as well. Let's take a quick peek

[01:35:44] well. Let's take a quick peek at one of the conversations I

[01:35:46] at one of the conversations I previously had.

[01:35:48] previously had. I was curious for this demo

[01:35:50] I was curious for this demo about some fun facts about Doom.

[01:35:51] about some fun facts about Doom. So I asked the agent to do some

[01:35:54] So I asked the agent to do some research. It generated and

[01:35:55] research. It generated and plots on the right side of the

[01:35:57] plots on the right side of the panel and then finally, it also

[01:35:58] panel and then finally, it also generated a cool artifact for

[01:35:59] generated a cool artifact for me.

[01:36:00] me. It even generated an infographic

[01:36:02] It even generated an infographic using Nano Banana Pro, it

[01:36:04] using Nano Banana Pro, it generated some graphs using some

[01:36:06] generated some graphs using some code it just wrote, and then

[01:36:07] code it just wrote, and then afterwards, it generated some

[01:36:11] afterwards, it generated some cool tables.

[01:36:12] cool tables. As you can see, Antigravity 2.0

[01:36:13] As you can see, Antigravity 2.0 is unabashedly agent first an

[01:36:15] is unabashedly agent first an Has been optimized to be the

[01:36:17] Has been optimized to be the best surface for you to interact

[01:36:19] best surface for you to interact with agents. Let's take a look

[01:36:21] with agents. Let's take a look at the previous conversation to

[01:36:21] at the previous conversation to see how it's going.

[01:36:23] see how it's going. Antigravity ended up doing a

[01:36:24] Antigravity ended up doing a whole host of research, ended up

[01:36:26] whole host of research, ended up writing over 100 lines of code

[01:36:28] writing over 100 lines of code and then finally the built the

[01:36:30] and then finally the built the operating system.

[01:36:31] operating system. Let's take a peek and see if it

[01:36:32] Let's take a peek and see if it works. Moment of truth.

[01:36:35] works. Moment of truth. Amazing.

[01:36:36] Amazing. [Applause]

[01:36:43] [Applause] That never gets old.

[01:36:47] That never gets old. While playing Doom on an

[01:36:48] While playing Doom on an operating system that

[01:36:49] operating system that Antigravity built is both fun

[01:36:50] Antigravity built is both fun and impressive, it hasn't

[01:36:54] and impressive, it hasn't stopped there. We've tasked the

[01:36:55] stopped there. We've tasked the agents to build a photo editing

[01:36:58] agents to build a photo editing suite, a realtime messaging app

[01:36:59] suite, a realtime messaging app and a multiuser collaboration

[01:37:00] and a multiuser collaboration platform, all with the same

[01:37:02] platform, all with the same results. Multiday engineering

[01:37:03] results. Multiday engineering efforts are collapsing into

[01:37:06] efforts are collapsing into hours, if not minutes.

[01:37:08] hours, if not minutes. This was made possible by the

[01:37:09] This was made possible by the new subagent teamwork

[01:37:10] new subagent teamwork capability.

[01:37:11] capability. We're excited to bring this to

[01:37:12] We're excited to bring this to you as an early research preview

[01:37:16] you as an early research preview in Antigravity.

[01:37:18] in Antigravity. [Applause]

[01:37:23] [Applause] Last but not least, 3.5 Flash is

[01:37:27] Last but not least, 3.5 Flash is incredibly fast.

[01:37:27] incredibly fast. Like Sundar said, it's four

[01:37:30] Like Sundar said, it's four times faster than other frontier

[01:37:32] times faster than other frontier models, but, as we know, agentic

[01:37:33] models, but, as we know, agentic coding is a token monster. So

[01:37:34] coding is a token monster. So we've taken it to another level

[01:37:36] we've taken it to another level in Antigravity. We've optimized

[01:37:38] in Antigravity. We've optimized Flash to be not just four times,

[01:37:44] Flash to be not just four times, but 12 times faster in

[01:37:45] but 12 times faster in Antigravity.

[01:37:47] Antigravity. [Applause]

[01:37:50] [Applause] And we're thrilled to give you

[01:37:51] And we're thrilled to give you all a taste of this experience,

[01:37:51] all a taste of this experience, starting today.

[01:37:53] starting today. What we showed you today isn't

[01:37:57] What we showed you today isn't just a vision; it's how we're

[01:37:58] just a vision; it's how we're building Antigravity to be the

[01:37:59] building Antigravity to be the most complete agentic

[01:38:02] most complete agentic development platform for

[01:38:03] development platform for everyone. We're doing it with

[01:38:05] everyone. We're doing it with the Google ecosystem, whether

[01:38:08] the Google ecosystem, whether it's integrating the tech stacks

[01:38:13] it's integrating the tech stacks and tools that you already use

[01:38:15] and tools that you already use or using Antigravity's Agent

[01:38:16] or using Antigravity's Agent Harness to power the next crop

[01:38:16] Harness to power the next crop of agentic experiences across

[01:38:17] of agentic experiences across Google products.

[01:38:18] Google products. Today, Antigravity 2.0 is

[01:38:19] Today, Antigravity 2.0 is available globally for everyone.

[01:38:22] available globally for everyone. Join us in the Developer Keynote

[01:38:22] Join us in the Developer Keynote as we demo all the new

[01:38:24] as we demo all the new capabilities.

[01:38:25] capabilities. Back to you, Sundar.

[01:38:26] Back to you, Sundar. [Applause]

[01:38:32] [Applause] >> SUNDAR PICHAI: Thanks, Varun.

[01:38:34] >> SUNDAR PICHAI: Thanks, Varun. It's incredible that Varun's

[01:38:36] It's incredible that Varun's entire OS was built by a team of

[01:38:39] entire OS was built by a team of subagents in just 12 hours for

[01:38:41] subagents in just 12 hours for such a low cost.

[01:38:42] such a low cost. What's amazing about Flash is

[01:38:45] What's amazing about Flash is how it delivers frontier-level

[01:38:47] how it delivers frontier-level capabilities at less than half

[01:38:49] capabilities at less than half the price of comparable frontier

[01:38:51] the price of comparable frontier models.

[01:38:52] models. We've heard that many companies

[01:38:54] We've heard that many companies are already blowing through

[01:38:57] are already blowing through their annual token budgets, and

[01:38:58] their annual token budgets, and it's only May.

[01:38:59] it's only May. If companies used a mix of Flash

[01:39:00] If companies used a mix of Flash and other frontier models, they

[01:39:01] and other frontier models, they could save a lot of money. To

[01:39:02] could save a lot of money. To put this in perspective, the top

[01:39:03] put this in perspective, the top companies in Google Cloud are

[01:39:04] companies in Google Cloud are processing about 1 trillion

[01:39:06] processing about 1 trillion tokens a day.

[01:39:10] tokens a day. If they shifted 80% of their

[01:39:13] If they shifted 80% of their workloads from other frontier

[01:39:16] workloads from other frontier models to 3.5 Flash, they would

[01:39:16] models to 3.5 Flash, they would save over $1 billion annually.

[01:39:20] save over $1 billion annually. Real savings they can pour back

[01:39:21] Real savings they can pour back into their company.

[01:39:27] into their company. Gemini 3.5 Flash is available

[01:39:28] Gemini 3.5 Flash is available for everyone today, across our

[01:39:28] for everyone today, across our products, and APIs.

[01:39:30] products, and APIs. [Applause]

[01:39:33] [Applause] We're also very excited for 3.5

[01:39:37] We're also very excited for 3.5 Pro. We are using it

[01:39:37] Pro. We are using it internally. It's showing great

[01:39:40] internally. It's showing great improvements. I know you can't

[01:39:41] improvements. I know you can't wait to get your hands on it.

[01:39:45] wait to get your hands on it. Give us until next month to get

[01:39:46] Give us until next month to get it to you.

[01:39:48] it to you. Gemini 3.5 and Antigravity are

[01:39:52] Gemini 3.5 and Antigravity are unlocking a new world of agents,

[01:39:53] unlocking a new world of agents, and agentic capabilities. We've

[01:39:54] and agentic capabilities. We've been bringing agents to

[01:39:55] been bringing agents to developers and enterprises for a

[01:39:59] developers and enterprises for a while. And now, we are super

[01:39:59] while. And now, we are super focused on bringing the power of

[01:40:02] focused on bringing the power of agents, safely and securely, to

[01:40:04] agents, safely and securely, to consumers, so that they work for

[01:40:07] consumers, so that they work for everyone.

[01:40:08] everyone. You will see many agentic

[01:40:14] You will see many agentic experiences across many of our

[01:40:15] experiences across many of our products today. I'm

[01:40:15] products today. I'm particularly excited for what we

[01:40:16] particularly excited for what we are bringing right into the

[01:40:17] are bringing right into the Gemini app.

[01:40:21] Gemini app. Introducing Gemini Spark.

[01:40:23] Introducing Gemini Spark. [Applause]

[01:40:27] [Applause] It's your personal AI agent that

[01:40:27] It's your personal AI agent that helps you navigate your digital

[01:40:29] helps you navigate your digital life, taking action on your

[01:40:32] life, taking action on your behalf and under your direction.

[01:40:36] behalf and under your direction. It runs on dedicated virtual

[01:40:38] It runs on dedicated virtual machines on Google Cloud, and it

[01:40:44] machines on Google Cloud, and it is 24/7. And yes, you can close

[01:40:46] is 24/7. And yes, you can close your laptop.

[01:40:48] your laptop. [Applause]

[01:40:49] [Applause] It's powered by Gemini 3.5, and

[01:40:51] It's powered by Gemini 3.5, and the Google Antigravity Harness,

[01:40:55] the Google Antigravity Harness, which allows it to perform

[01:40:55] which allows it to perform long-running tasks easily in the

[01:40:58] long-running tasks easily in the background. Spark integrates

[01:41:00] background. Spark integrates seamlessly with tools, starting

[01:41:01] seamlessly with tools, starting with our own, and in the coming

[01:41:03] with our own, and in the coming weeks, with third-party tools

[01:41:05] weeks, with third-party tools through MCP.

[01:41:07] through MCP. And you can work with Spark

[01:41:10] And you can work with Spark however is most convenient, in

[01:41:11] however is most convenient, in the Gemini app or soon through

[01:41:13] the Gemini app or soon through e-mail and chat.

[01:41:14] e-mail and chat. Let's have Josh come up and

[01:41:15] Let's have Josh come up and share more.

[01:41:16] share more. ♪

[01:41:17] ♪ ♪

[01:41:19] ♪ [Applause]

[01:41:29] [Applause] >> JOSH WOODWARD: Thanks,

[01:41:30] >> JOSH WOODWARD: Thanks, Sundar. It's great to see

[01:41:31] Sundar. It's great to see everyone and let me show you how

[01:41:32] everyone and let me show you how Spark works with some examples

[01:41:33] Spark works with some examples from my personal life.

[01:41:35] from my personal life. So here we are, we have the new

[01:41:39] So here we are, we have the new Gemini open, it's been

[01:41:40] Gemini open, it's been completely redesigned. We'll

[01:41:41] completely redesigned. We'll talk about that later in the

[01:41:42] talk about that later in the show, and I want take you into

[01:41:44] show, and I want take you into Spark here and what you can see

[01:41:44] Spark here and what you can see immediately is a dashboard with

[01:41:45] immediately is a dashboard with all the different tasks that I

[01:41:46] all the different tasks that I have going on in the background.

[01:41:47] have going on in the background. It allows you to check in on

[01:41:48] It allows you to check in on these things, and I'll paste in

[01:41:50] these things, and I'll paste in a task right off the bat. This

[01:41:52] a task right off the bat. This is a pretty straightforward

[01:41:53] is a pretty straightforward example, but it's so useful.

[01:41:54] example, but it's so useful. Help me draft an e-mail to the

[01:41:58] Help me draft an e-mail to the team, compile everything about

[01:41:59] team, compile everything about our recent Gemini live launches

[01:42:01] our recent Gemini live launches and wins from the last week.

[01:42:04] and wins from the last week. Use slash ghost writer so

[01:42:05] Use slash ghost writer so there's a few things happening

[01:42:07] there's a few things happening here. Compile everything. This

[01:42:09] here. Compile everything. This goes across Docs, e-mail, your

[01:42:12] goes across Docs, e-mail, your chats and grabs the most

[01:42:14] chats and grabs the most important information that you

[01:42:14] important information that you need for this update. It also

[01:42:17] need for this update. It also is going to use all of the stuff

[01:42:19] is going to use all of the stuff you see in the time period of

[01:42:21] you see in the time period of the last week with the slash

[01:42:22] the last week with the slash ghost writer. That's a personal

[01:42:24] ghost writer. That's a personal skill that I've written so the

[01:42:26] skill that I've written so the e-mail sounds like me.

[01:42:27] e-mail sounds like me. And what's great is that with

[01:42:29] And what's great is that with Spark, you can upload your

[01:42:31] Spark, you can upload your favorite skills you find online.

[01:42:33] favorite skills you find online. So we're going to let this run

[01:42:34] So we're going to let this run in the background. You can see

[01:42:36] in the background. You can see it's already started doing

[01:42:38] it's already started doing various tool calls, and I'm

[01:42:39] various tool calls, and I'm going to switch over to another

[01:42:40] going to switch over to another one from our personal life.

[01:42:41] one from our personal life. We're planning a big block

[01:42:43] We're planning a big block party, and you can see here,

[01:42:43] party, and you can see here, this is a pretty complex prompt

[01:42:46] this is a pretty complex prompt We want help grabbing all the

[01:42:48] We want help grabbing all the RSVPs, keep a list of who's

[01:42:51] RSVPs, keep a list of who's bringing what, remember to

[01:42:52] bringing what, remember to e-mail those neighbors who

[01:42:53] e-mail those neighbors who haven't signed up yet.

[01:42:55] haven't signed up yet. And what's amazing here is Spark

[01:42:57] And what's amazing here is Spark will go through, step by step,

[01:43:00] will go through, step by step, look at all these steps, all the

[01:43:01] look at all these steps, all the time it saves you going through

[01:43:04] time it saves you going through and again work across the

[01:43:05] and again work across the various skills and apps that you

[01:43:07] various skills and apps that you have.

[01:43:07] have. And what's really amazing is it

[01:43:09] And what's really amazing is it will break it down and also be

[01:43:11] will break it down and also be able to generate files for you.

[01:43:12] able to generate files for you. So the first one here, this is a

[01:43:14] So the first one here, this is a live RSVP tracker, right in

[01:43:18] live RSVP tracker, right in Google Sheets. You can see that

[01:43:19] Google Sheets. You can see that it shows who's confirmed and who

[01:43:21] it shows who's confirmed and who hasn't.

[01:43:21] hasn't. What's amazing about this is it

[01:43:23] What's amazing about this is it will actually update, because

[01:43:24] will actually update, because it's connected to Gmail. So

[01:43:27] it's connected to Gmail. So when L. Thompson wrote 8RSVPs,

[01:43:30] when L. Thompson wrote 8RSVPs, it will update, which is pretty

[01:43:32] it will update, which is pretty amazing.

[01:43:33] amazing. The other thing is it keeps

[01:43:34] The other thing is it keeps track of all the different

[01:43:37] track of all the different guests, as well as sending

[01:43:38] guests, as well as sending follow-up reminders to who

[01:43:38] follow-up reminders to who hasn't signed up yet.

[01:43:40] hasn't signed up yet. So again, this will happen. It

[01:43:41] So again, this will happen. It will create the drafts and under

[01:43:43] will create the drafts and under my control, I can send them.

[01:43:46] my control, I can send them. And then finally what the prompt

[01:43:47] And then finally what the prompt did is it created a hype deck

[01:43:50] did is it created a hype deck for the block party and you can

[01:43:51] for the block party and you can see it creates right in Google

[01:43:54] see it creates right in Google Slides, perfectly integrated and

[01:43:55] Slides, perfectly integrated and you can see it even pulls in

[01:43:56] you can see it even pulls in things like our giant bounce

[01:43:57] things like our giant bounce house that's going to be in the

[01:43:59] house that's going to be in the cul-de-sac.

[01:43:59] cul-de-sac. Now, all of this is happening in

[01:44:01] Now, all of this is happening in the background and under my

[01:44:05] the background and under my control.

[01:44:05] control. And what's amazing is Gemini can

[01:44:07] And what's amazing is Gemini can even go a step further and pull

[01:44:09] even go a step further and pull out things like your

[01:44:10] out things like your neighborhood homeowners

[01:44:11] neighborhood homeowners association won't let you set

[01:44:12] association won't let you set this up before Friday afternoon

[01:44:14] this up before Friday afternoon on June 5th. That's pulling

[01:44:15] on June 5th. That's pulling from a file in my Google Drive.

[01:44:18] from a file in my Google Drive. So it's incredibly helpful about

[01:44:18] So it's incredibly helpful about how it pulls it all together.

[01:44:22] how it pulls it all together. Now, that's Spark working on a

[01:44:24] Now, that's Spark working on a laptop.

[01:44:24] laptop. Spark is also amazing on the go,

[01:44:27] Spark is also amazing on the go, and it works across both Android

[01:44:28] and it works across both Android and iPhone. So here you can

[01:44:30] and iPhone. So here you can see on my phone here I'll open

[01:44:32] see on my phone here I'll open it up, and I can go into Spark.

[01:44:35] it up, and I can go into Spark. You see both of the tasks we

[01:44:36] You see both of the tasks we just went through, they sync

[01:44:38] just went through, they sync across all of your devices,

[01:44:40] across all of your devices, which is so helpful and Spark is

[01:44:41] which is so helpful and Spark is just amazing at brain-dumping

[01:44:42] just amazing at brain-dumping things on your mind.

[01:44:44] things on your mind. If you're super busy, it's

[01:44:46] If you're super busy, it's almost, you can just throw tasks

[01:44:47] almost, you can just throw tasks over your shoulder.

[01:44:48] over your shoulder. Spark will catch them and then

[01:44:50] Spark will catch them and then run with them. So watch this.

[01:44:53] run with them. So watch this. Start a few threads for me. The

[01:44:55] Start a few threads for me. The first one, find all the upcoming

[01:44:57] first one, find all the upcoming meetings with Sundar and turn

[01:44:58] meetings with Sundar and turn them all hot pink so I don't

[01:45:00] them all hot pink so I don't miss them.

[01:45:04] miss them. Write The second one last nightt

[01:45:10] Write The second one last nightt our new neighbor, John. Write a

[01:45:13] our new neighbor, John. Write a note to him and his him and his

[01:45:16] note to him and his him and his family, invite them to our block

[01:45:18] family, invite them to our block party, because they weren't on

[01:45:19] party, because they weren't on our list originally. The third

[01:45:20] our list originally. The third one, create a document with the

[01:45:22] one, create a document with the top things my wife, and I need

[01:45:23] top things my wife, and I need to do for the kids before the

[01:45:24] to do for the kids before the end of the school year.

[01:45:26] end of the school year. Categorize it by deadline and

[01:45:28] Categorize it by deadline and priority and make it easy to

[01:45:29] priority and make it easy to digest. I don't want to miss

[01:45:30] digest. I don't want to miss anything. All right. We'll

[01:45:31] anything. All right. We'll send that in and you can see at

[01:45:32] send that in and you can see at the speed of my voice it's

[01:45:33] the speed of my voice it's taking that one task, and it

[01:45:35] taking that one task, and it will capture all of that context

[01:45:37] will capture all of that context as fast as I can talk.

[01:45:38] as fast as I can talk. It starts out as a single thread

[01:45:40] It starts out as a single thread here and in the background it's

[01:45:41] here and in the background it's actually going to go through and

[01:45:43] actually going to go through and break down those into individual

[01:45:45] break down those into individual tasks. Now, I can just put my

[01:45:46] tasks. Now, I can just put my phone away and get on with my

[01:45:49] phone away and get on with my day. And Spark works in the

[01:45:50] day. And Spark works in the background for me. We'll check

[01:45:51] background for me. We'll check in later and see how it's doing.

[01:45:55] in later and see how it's doing. You can see this is one of the

[01:45:57] You can see this is one of the first times we've been able to

[01:45:58] first times we've been able to put a phone down and let it keep

[01:46:00] put a phone down and let it keep working on the I/O stage. It's

[01:46:02] working on the I/O stage. It's great.

[01:46:03] great. Because we're prioritizing

[01:46:03] Because we're prioritizing safety, we're rolling out Gemini

[01:46:06] safety, we're rolling out Gemini Spark deliberately, to trusted

[01:46:08] Spark deliberately, to trusted testers, this week, and as a

[01:46:10] testers, this week, and as a beta for U.S. Google AI Ultra

[01:46:13] beta for U.S. Google AI Ultra subscribers next week. We want

[01:46:15] subscribers next week. We want this new type of help to be in

[01:46:17] this new type of help to be in as many hands as possible, so to

[01:46:19] as many hands as possible, so to do that, we're introducing a new

[01:46:24] do that, we're introducing a new Ultra plan for $100 a month.

[01:46:28] Ultra plan for $100 a month. [Applause]

[01:46:32] [Applause] And for those of you that need

[01:46:33] And for those of you that need maximum limits, we're dropping

[01:46:35] maximum limits, we're dropping the price for our top tier Ultra

[01:46:37] the price for our top tier Ultra plan from $250 a month to $200 a

[01:46:40] plan from $250 a month to $200 a month.

[01:46:44] month. And there's so much more to

[01:46:45] And there's so much more to come. Later, this summer,

[01:46:46] come. Later, this summer, Gemini Spark will operate

[01:46:49] Gemini Spark will operate directly within Chrome, acting

[01:46:50] directly within Chrome, acting as your agentic browser across

[01:46:53] as your agentic browser across the web. It can take action on

[01:46:55] the web. It can take action on your tasks under your direction.

[01:46:58] your tasks under your direction. We're also building a dedicated

[01:46:59] We're also building a dedicated home base for your agents on

[01:47:01] home base for your agents on your phone, Android Halo, which

[01:47:05] your phone, Android Halo, which is coming later this year.

[01:47:05] is coming later this year. As Sundar said, we've entered a

[01:47:08] As Sundar said, we've entered a new agentic era across Google,

[01:47:10] new agentic era across Google, and we can't wait to see what

[01:47:11] and we can't wait to see what you're going to build with it.

[01:47:12] you're going to build with it. Back to you, Sundar.

[01:47:15] Back to you, Sundar. [Applause]

[01:47:18] [Applause] >> SUNDAR PICHAI: Thanks, Josh.

[01:47:21] >> SUNDAR PICHAI: Thanks, Josh. So great to see Gemini Spark

[01:47:23] So great to see Gemini Spark getting things done on your

[01:47:26] getting things done on your behalf. I've played around with

[01:47:27] behalf. I've played around with all sorts of agents and you can

[01:47:28] all sorts of agents and you can really see the potential. Still

[01:47:31] really see the potential. Still early days when it comes to

[01:47:32] early days when it comes to making agents easy to use, super

[01:47:37] making agents easy to use, super secure, and truly helpful.

[01:47:37] secure, and truly helpful. That's why I'm really excited by

[01:47:38] That's why I'm really excited by Gemini Spark. We are laying the

[01:47:39] Gemini Spark. We are laying the foundation to bring this all

[01:47:41] foundation to bring this all together in a safe and secure

[01:47:43] together in a safe and secure way to consumers everywhere. We

[01:47:44] way to consumers everywhere. We look forward to having you all

[01:47:46] look forward to having you all try it.

[01:47:48] try it. We are firmly in our agentic

[01:47:51] We are firmly in our agentic Gemini era. Gemini Spark is the

[01:47:54] Gemini era. Gemini Spark is the first experience you're seeing,

[01:47:55] first experience you're seeing, made possible by 3.5 models and

[01:47:57] made possible by 3.5 models and Antigravity. This combination

[01:48:00] Antigravity. This combination gives us new ways to actuate our

[01:48:01] gives us new ways to actuate our mission and transform our

[01:48:03] mission and transform our products to be radically more

[01:48:05] products to be radically more helpful.

[01:48:07] helpful. I can't wait to see how it will

[01:48:09] I can't wait to see how it will transform Search, our ultimate

[01:48:11] transform Search, our ultimate moonshot. This past year has

[01:48:14] moonshot. This past year has proven just how much innovation

[01:48:16] proven just how much innovation is possible right at the heart

[01:48:21] is possible right at the heart of our information mission.

[01:48:22] of our information mission. As we enter this agentic era,

[01:48:22] As we enter this agentic era, Search will be more helpful and

[01:48:23] Search will be more helpful and powerful than ever before.

[01:48:25] powerful than ever before. Let me turn it over to Liz to

[01:48:26] Let me turn it over to Liz to share what's next.

[01:48:27] share what's next. ♪

[01:48:29] ♪ ♪

[01:48:30] ♪ [Applause]

[01:48:40] [Applause] >> LIZ REID: People bring

[01:48:42] >> LIZ REID: People bring billions of questions to Search

[01:48:47] billions of questions to Search every day. Now sometimes, the

[01:48:48] every day. Now sometimes, the whole world is searching for the

[01:48:50] whole world is searching for the same thing; but more often, your

[01:48:51] same thing; but more often, your questions are just as unique as

[01:48:56] questions are just as unique as you are. that's why we set out

[01:48:57] you are. that's why we set out to make it possible to ask

[01:48:58] to make it possible to ask whatever is really on your mind.

[01:49:02] whatever is really on your mind. To unlock this, we've been on a

[01:49:03] To unlock this, we've been on a journey to bring together the

[01:49:04] journey to bring together the best of a search engine with the

[01:49:05] best of a search engine with the best of AI. We started this

[01:49:09] best of AI. We started this transformation with AI Overview.

[01:49:10] transformation with AI Overview. And if you can believe it, it

[01:49:11] And if you can believe it, it was just last year on this stage

[01:49:13] was just last year on this stage that we launched AI Mode.

[01:49:15] that we launched AI Mode. It's our most powerful AI

[01:49:17] It's our most powerful AI Search, bringing in our most

[01:49:20] Search, bringing in our most advanced Gemini models. And as

[01:49:21] advanced Gemini models. And as of today, we're upgrading it on

[01:49:21] of today, we're upgrading it on Gemini 3.5.

[01:49:27] Gemini 3.5. [Applause]

[01:49:32] [Applause] Now, as Sundar mentioned, AI

[01:49:32] Now, as Sundar mentioned, AI Mode has surpassed more than 1

[01:49:33] Mode has surpassed more than 1 billion monthly users, and we're

[01:49:36] billion monthly users, and we're seeing phenomenal growth, with

[01:49:37] seeing phenomenal growth, with AI Mode queries more than

[01:49:40] AI Mode queries more than doubling every quarter since

[01:49:42] doubling every quarter since launch. As people have learned

[01:49:43] launch. As people have learned how much more Search can do,

[01:49:44] how much more Search can do, they started bringing more

[01:49:48] they started bringing more questions, so much so that last

[01:49:51] questions, so much so that last quarter, we saw Search queries

[01:49:52] quarter, we saw Search queries reach an all-time high.

[01:49:55] reach an all-time high. reach an all-time high.

[01:49:56] reach an all-time high. But what's even more remarkable:

[01:49:57] But what's even more remarkable: You're asking your real

[01:50:00] You're asking your real questions in all their super

[01:50:02] questions in all their super specific detailed glory, knowing

[01:50:03] specific detailed glory, knowing Search can really tackle them.

[01:50:04] Search can really tackle them. You're having real conversations

[01:50:08] You're having real conversations with Search, going back and

[01:50:08] with Search, going back and forth, and deeper. So you're

[01:50:13] forth, and deeper. So you're not just asking nearby hikes.

[01:50:13] not just asking nearby hikes. You're asking can you build an

[01:50:15] You're asking can you build an itinerary for hiking day trip

[01:50:16] itinerary for hiking day trip near me with great views and dog

[01:50:17] near me with great views and dog friendly trails, and a lunch

[01:50:18] friendly trails, and a lunch spot with convenient parking?

[01:50:21] spot with convenient parking? And now, we're entering the next

[01:50:23] And now, we're entering the next chapter of Google Search, where

[01:50:26] chapter of Google Search, where incredible AI features aren't

[01:50:29] incredible AI features aren't just in Search; Google Search is

[01:50:32] just in Search; Google Search is AI Search through and through.

[01:50:35] AI Search through and through. Now, it's an AI Search that

[01:50:36] Now, it's an AI Search that brings to figure out our most

[01:50:37] brings to figure out our most advanced Gemini models, our

[01:50:39] advanced Gemini models, our newest agentic capabilities, and

[01:50:39] newest agentic capabilities, and the full breadth of the world's

[01:50:40] the full breadth of the world's information.

[01:50:40] information. With over 1 billion facts

[01:50:46] With over 1 billion facts updated a minute, billions of

[01:50:47] updated a minute, billions of new web pages indexed every day,

[01:50:48] new web pages indexed every day, and connections to an infinite

[01:50:50] and connections to an infinite range of human perspectives.

[01:50:51] range of human perspectives. So whatever is on your mind, you

[01:50:54] So whatever is on your mind, you can come to Google and truly ask

[01:50:55] can come to Google and truly ask anything.

[01:50:59] anything. Now, to start, I'm excited to

[01:51:00] Now, to start, I'm excited to announce we're launching a

[01:51:02] announce we're launching a brand-new intelligent Search

[01:51:04] brand-new intelligent Search box. Before, the Search box was

[01:51:07] box. Before, the Search box was a contained space, but now, it's

[01:51:11] a contained space, but now, it's totally reimagined with AI. It

[01:51:11] totally reimagined with AI. It expands with your curiosity. As

[01:51:12] expands with your curiosity. As you ask, Search helps you

[01:51:18] you ask, Search helps you formulate your question with

[01:51:18] formulate your question with AI-powered suggestions.

[01:51:20] AI-powered suggestions. This goes beyond autocomplete.

[01:51:21] This goes beyond autocomplete. It offers nuances that you might

[01:51:22] It offers nuances that you might not have even thought to add,

[01:51:24] not have even thought to add, helping you take the exact

[01:51:25] helping you take the exact question on your mind and ask it

[01:51:26] question on your mind and ask it with ease.

[01:51:28] with ease. This new Search box puts our

[01:51:30] This new Search box puts our most powerful AI tools right at

[01:51:34] most powerful AI tools right at your fingertips. And you can

[01:51:35] your fingertips. And you can ask across modalities with text,

[01:51:38] ask across modalities with text, images, files, and videos, and

[01:51:39] images, files, and videos, and Search reasons across them all.

[01:51:41] Search reasons across them all. This is the biggest upgrade to

[01:51:44] This is the biggest upgrade to our iconic Search box since its

[01:51:47] our iconic Search box since its debut over 25 years ago. And

[01:51:47] debut over 25 years ago. And it's starting to roll out toda

[01:51:51] it's starting to roll out toda [Applause]

[01:51:59] [Applause] Next, we're making it even

[01:51:59] Next, we're making it even easier to continue the

[01:52:00] easier to continue the conversation with Search,

[01:52:01] conversation with Search, bringing AI Overviews and AI

[01:52:02] bringing AI Overviews and AI Mode into one seamless AI Search

[01:52:03] Mode into one seamless AI Search experience. You can float

[01:52:06] experience. You can float effortlessly from your question

[01:52:13] effortlessly from your question to your response on the main

[01:52:13] to your response on the main Search results page, to

[01:52:17] Search results page, to follow-ups in AI Mode. Your

[01:52:18] follow-ups in AI Mode. Your context stays with you and your

[01:52:18] context stays with you and your conversation gets deeper. Your

[01:52:19] conversation gets deeper. Your links and sources get even more

[01:52:21] links and sources get even more relevant to what you might want

[01:52:23] relevant to what you might want to explore, so you continue to

[01:52:23] to explore, so you continue to get the best of AI and the best

[01:52:24] get the best of AI and the best of the Web.

[01:52:25] of the Web. And I'm excited to share this

[01:52:26] And I'm excited to share this new seamless AI Search

[01:52:28] new seamless AI Search experience is live today, across

[01:52:30] experience is live today, across desktop and mobile worldwide.

[01:52:32] desktop and mobile worldwide. [Applause]

[01:52:39] [Applause] Next, you just heard Sundar and

[01:52:40] Next, you just heard Sundar and Josh share our approach to

[01:52:43] Josh share our approach to agents and the potential that

[01:52:46] agents and the potential that this can open up. Now, we're

[01:52:47] this can open up. Now, we're taking an exciting step toward

[01:52:50] taking an exciting step toward this vision, where you'll be

[01:52:50] this vision, where you'll be able to create and manage

[01:52:51] able to create and manage multiple AI agents for your many

[01:52:51] multiple AI agents for your many tasks, right in Search.

[01:52:55] tasks, right in Search. We're entering the era of Search

[01:52:59] We're entering the era of Search agents. Now to start, you can

[01:52:59] agents. Now to start, you can set information agents to work

[01:53:01] set information agents to work for you 24/7 in the background.

[01:53:04] for you 24/7 in the background. They can find you exactly what

[01:53:05] They can find you exactly what you need, exactly when you need

[01:53:06] you need, exactly when you need it, and help you take action.

[01:53:09] it, and help you take action. You can spin up multiple agents

[01:53:13] You can spin up multiple agents in Search simultaneously, so you

[01:53:14] in Search simultaneously, so you can get updated and make

[01:53:14] can get updated and make progress on all the things that

[01:53:17] progress on all the things that matter to you.

[01:53:19] matter to you. And these will work with and

[01:53:20] And these will work with and alongside Gemini Spark to help

[01:53:22] alongside Gemini Spark to help you get more done, so let's put

[01:53:23] you get more done, so let's put a few to work.

[01:53:23] a few to work. Say you're really into finance,

[01:53:24] Say you're really into finance, and you want to know more about

[01:53:27] and you want to know more about big biotech stocks with P/E

[01:53:31] big biotech stocks with P/E under 15, positive cash flow and

[01:53:32] under 15, positive cash flow and low debt, right when it matters.

[01:53:36] low debt, right when it matters. Now, you can just ask all of

[01:53:36] Now, you can just ask all of that and your agent is off. It

[01:53:38] that and your agent is off. It takes your super complex

[01:53:41] takes your super complex question and map out a plan.

[01:53:42] question and map out a plan. It determines the urgency,

[01:53:46] It determines the urgency, getting that you really need

[01:53:47] getting that you really need in-the-moment intel, and it sets

[01:53:48] in-the-moment intel, and it sets triggers to look out for

[01:53:51] triggers to look out for information as it changes, and

[01:53:51] information as it changes, and picks the tools and data hooks

[01:53:53] picks the tools and data hooks it needs for the job.

[01:53:53] it needs for the job. It connects directly to our

[01:53:55] It connects directly to our realtime finance data so you get

[01:53:58] realtime finance data so you get up-to-the-second updates on

[01:53:59] up-to-the-second updates on stock prices and insights on the

[01:54:03] stock prices and insights on the market, the moment it moves.

[01:54:07] market, the moment it moves. Now, when it does, your agent

[01:54:08] Now, when it does, your agent sends you an intelligent and

[01:54:09] sends you an intelligent and synthesized update.

[01:54:09] synthesized update. It helps you understand what's

[01:54:10] It helps you understand what's going on so you can separate the

[01:54:13] going on so you can separate the signal from the noise. And it

[01:54:14] signal from the noise. And it points you to hyper-relevant

[01:54:16] points you to hyper-relevant content, like this crowdsourced

[01:54:18] content, like this crowdsourced research platform, news site and

[01:54:22] research platform, news site and social.

[01:54:23] social. This helps websites and creators

[01:54:23] This helps websites and creators get fresh content discovered by

[01:54:25] get fresh content discovered by people who really care about it,

[01:54:26] people who really care about it, when it matters most to them.

[01:54:31] when it matters most to them. Now, let's say you're apartment

[01:54:32] Now, let's say you're apartment hunting. You can do a total

[01:54:32] hunting. You can do a total brain-dump of what you're

[01:54:35] brain-dump of what you're looking for, with all your

[01:54:36] looking for, with all your criteria like location and

[01:54:38] criteria like location and natural light and availability,

[01:54:39] natural light and availability, and your agent continuously

[01:54:41] and your agent continuously scans the entire web, across

[01:54:45] scans the entire web, across sites, social, and forums.

[01:54:46] sites, social, and forums. Or if you're a sneaker fan, you

[01:54:46] Or if you're a sneaker fan, you can just ask to be updated when

[01:54:47] can just ask to be updated when any of your favorite athletes

[01:54:52] any of your favorite athletes announce sneaker collabs or

[01:54:52] announce sneaker collabs or drops.

[01:54:56] drops. And it monitors everything from

[01:54:56] And it monitors everything from blogs to our Shopping Graph so

[01:54:57] blogs to our Shopping Graph so you don't miss out.

[01:54:58] you don't miss out. You'll be able to put

[01:54:58] You'll be able to put information agents to work for

[01:55:01] information agents to work for you this summer. Just ask

[01:55:02] you this summer. Just ask Search to keep you updated on

[01:55:04] Search to keep you updated on whatever you want to know.

[01:55:06] whatever you want to know. Now, information agents are

[01:55:12] Now, information agents are among the first of many agents

[01:55:12] among the first of many agents we're introducing in Search to

[01:55:13] we're introducing in Search to make it more helpful for you.

[01:55:14] make it more helpful for you. So whether you want to find it,

[01:55:15] So whether you want to find it, or check it, book it, buy it or

[01:55:17] or check it, book it, buy it or more, Search will be able to get

[01:55:17] more, Search will be able to get it done.

[01:55:23] it done. Now, we're also bringing agentic

[01:55:25] Now, we're also bringing agentic coding to Search so it can build

[01:55:26] coding to Search so it can build you custom experiences just for

[01:55:29] you custom experiences just for your questions. To show you how

[01:55:30] your questions. To show you how this works, here's Robby.

[01:55:30] this works, here's Robby. ♪

[01:55:30] ♪ ♪

[01:55:31] ♪ [Applause]

[01:55:41] [Applause] >> ROBBY STEIN: We believe the

[01:55:46] >> ROBBY STEIN: We believe the best version of Search is one

[01:55:47] best version of Search is one created just for you. It's a

[01:55:50] created just for you. It's a Search that gives you

[01:55:51] Search that gives you information in the most helpful

[01:55:54] information in the most helpful format for your question. And

[01:55:55] format for your question. And we've spent years perfecting

[01:55:57] we've spent years perfecting this.

[01:55:57] this. So if you're shopping, we give

[01:55:58] So if you're shopping, we give you products. Asking about

[01:56:01] you products. Asking about data, you see charts. Looking

[01:56:04] data, you see charts. Looking for inspiration, you get

[01:56:07] for inspiration, you get beautiful visuals.

[01:56:09] beautiful visuals. Now, we're taking this to a

[01:56:10] Now, we're taking this to a whole new level. We're bringing

[01:56:12] whole new level. We're bringing Antigravity and the agentic

[01:56:15] Antigravity and the agentic coding capabilities of Gemini

[01:56:18] coding capabilities of Gemini 3.5 Flash right into Search.

[01:56:20] 3.5 Flash right into Search. So Search can build you the

[01:56:22] So Search can build you the ideal format exactly for your

[01:56:24] ideal format exactly for your question, completely custom on

[01:56:25] question, completely custom on the fly.

[01:56:29] the fly. We're talking dynamic layouts,

[01:56:31] We're talking dynamic layouts, interactive widgets, entire

[01:56:35] interactive widgets, entire experiences, all created just

[01:56:39] experiences, all created just for you. This is agentic coding

[01:56:40] for you. This is agentic coding at the scale of Search.

[01:56:47] at the scale of Search. [Applause]

[01:56:52] [Applause] Let me give you an example. Say

[01:56:53] Let me give you an example. Say I'm a college student trying to

[01:56:54] I'm a college student trying to wrap my mind around

[01:56:54] wrap my mind around astrophysics. I can go to

[01:56:56] astrophysics. I can go to Search and just ask how do black

[01:56:57] Search and just ask how do black holes affect space-time? And

[01:56:58] holes affect space-time? And check this out. I get an

[01:56:59] check this out. I get an interactive visual right in the

[01:57:03] interactive visual right in the AI Overview. Search now gets

[01:57:04] AI Overview. Search now gets that for a concept this complex,

[01:57:09] that for a concept this complex, I need to interact with it to

[01:57:10] I need to interact with it to really understand it.

[01:57:13] really understand it. This is still kind of 101, so

[01:57:15] This is still kind of 101, so I'm going to follow up. Show me

[01:57:16] I'm going to follow up. Show me how two orbiting objects like

[01:57:17] how two orbiting objects like binary black holes create

[01:57:22] binary black holes create gravitational waves. Search

[01:57:23] gravitational waves. Search dynamically builds a brand-new

[01:57:24] dynamically builds a brand-new interactive visual in real time

[01:57:28] interactive visual in real time completely custom for my

[01:57:29] completely custom for my specific question.

[01:57:31] specific question. [Applause]

[01:57:37] [Applause] And now, I can play with the

[01:57:37] And now, I can play with the parameters now like orbital

[01:57:38] parameters now like orbital separation, and the mass ratio,

[01:57:41] separation, and the mass ratio, and it's so cool. I can see we

[01:57:46] and it's so cool. I can see we how wave patterns have changed,

[01:57:47] how wave patterns have changed, and now, the smaller black hole

[01:57:50] and now, the smaller black hole spirals around the bigger one.

[01:57:50] spirals around the bigger one. And now that I get the basics, I

[01:57:51] And now that I get the basics, I can dive into these resources

[01:57:52] can dive into these resources here and read things like the

[01:57:54] here and read things like the LIGO Discovery Papers to learn

[01:57:54] LIGO Discovery Papers to learn more.

[01:57:58] more. Now, you might be wondering how

[01:58:00] Now, you might be wondering how can Search actually build custom

[01:58:01] can Search actually build custom UI like this for billions of

[01:58:03] UI like this for billions of unique questions?

[01:58:05] unique questions? With Gemini 3.5 Flash, Search

[01:58:08] With Gemini 3.5 Flash, Search plans the ideal response from

[01:58:12] plans the ideal response from scratch. It designs the layout,

[01:58:13] scratch. It designs the layout, decides what custom components

[01:58:14] decides what custom components to build, fans out to research

[01:58:19] to build, fans out to research and then finally deploys the

[01:58:19] and then finally deploys the code. To build custom

[01:58:20] code. To build custom components in the response like

[01:58:22] components in the response like this, Search invokes an agentic

[01:58:23] this, Search invokes an agentic coding harness powered by

[01:58:27] coding harness powered by Antigravity, so it can read and

[01:58:28] Antigravity, so it can read and write files and execute code in

[01:58:29] write files and execute code in a secure, containerized

[01:58:31] a secure, containerized environment.

[01:58:32] environment. This is the tech you saw Varun

[01:58:36] This is the tech you saw Varun build a whole OS with, and we're

[01:58:36] build a whole OS with, and we're bringing that power right into

[01:58:38] bringing that power right into Search.

[01:58:43] Search. Generative UI with Antigravity

[01:58:44] Generative UI with Antigravity is rolling out to Search this

[01:58:44] is rolling out to Search this summer, for everyone, free of

[01:58:45] summer, for everyone, free of charge.

[01:58:47] charge. [Applause]

[01:58:52] [Applause] So whatever you want to

[01:58:53] So whatever you want to understand, whether you're

[01:58:54] understand, whether you're wondering how your watch

[01:58:56] wondering how your watch actually works, analyzing the

[01:58:58] actually works, analyzing the cost of a new commute, or so

[01:59:03] cost of a new commute, or so much more, you'll get responses

[01:59:06] much more, you'll get responses as unique as your questions.

[01:59:07] as unique as your questions. Now, let's take this a step

[01:59:08] Now, let's take this a step further. Some projects aren't

[01:59:08] further. Some projects aren't one-off questions; they're

[01:59:11] one-off questions; they're ongoing tasks. So now, Search

[01:59:15] ongoing tasks. So now, Search has the ability to help you

[01:59:16] has the ability to help you build entire custom stateful

[01:59:22] build entire custom stateful experiences: Tools, trackers,

[01:59:25] experiences: Tools, trackers, dashboards. I think of these

[01:59:25] dashboards. I think of these sort of like building my own

[01:59:26] sort of like building my own little mini apps in Search, and

[01:59:26] little mini apps in Search, and I've been making a ton of them,

[01:59:27] I've been making a ton of them, and they're especially awesome

[01:59:30] and they're especially awesome for those long-running tasks

[01:59:30] for those long-running tasks where you want to keep coming

[01:59:32] where you want to keep coming back, like planning a wedding,

[01:59:35] back, like planning a wedding, or managing your home move.

[01:59:36] or managing your home move. So what do you think? Should we

[01:59:37] So what do you think? Should we build one together?

[01:59:40] build one together? Yeah?

[01:59:40] Yeah? [Applause]

[01:59:41] [Applause] All right.

[01:59:42] All right. Let's do a demo.

[01:59:44] Let's do a demo. So I'm constantly trying to

[01:59:46] So I'm constantly trying to figure out on Thursday what it

[01:59:47] figure out on Thursday what it is I should do with my family

[01:59:49] is I should do with my family every weekend.

[01:59:50] every weekend. So here's the search I actually

[01:59:52] So here's the search I actually just did about fun things to do

[01:59:53] just did about fun things to do with my family this weekend and,

[01:59:56] with my family this weekend and, of course, you have a great

[01:59:57] of course, you have a great response here from AI Mode, but

[01:59:59] response here from AI Mode, but here's something new.

[02:00:02] here's something new. Search proactively offers to

[02:00:03] Search proactively offers to build a weekend planner for me

[02:00:04] build a weekend planner for me and similar to how you just saw

[02:00:07] and similar to how you just saw Search create these generative

[02:00:08] Search create these generative UI and interactive visuals

[02:00:10] UI and interactive visuals completely from scratch, Search

[02:00:11] completely from scratch, Search can code this up right now.

[02:00:13] can code this up right now. Let's do it.

[02:00:14] Let's do it. So for the sake of today's demo,

[02:00:16] So for the sake of today's demo, we thought you might want to see

[02:00:17] we thought you might want to see a peek under the hood so as this

[02:00:19] a peek under the hood so as this builds, you're going to see

[02:00:21] builds, you're going to see realtime thinking steps and code

[02:00:22] realtime thinking steps and code generation flowing in.

[02:00:23] generation flowing in. And here, Search is thinking

[02:00:25] And here, Search is thinking about the right components to

[02:00:26] about the right components to build, not just what information

[02:00:28] build, not just what information to fetch, but the best way to

[02:00:30] to fetch, but the best way to present it, and I've chosen to

[02:00:32] present it, and I've chosen to securely connect Gmail, Photos

[02:00:34] securely connect Gmail, Photos and Calendar so it's using

[02:00:35] and Calendar so it's using personal intelligence to make my

[02:00:38] personal intelligence to make my suggestions even more helpful.

[02:00:38] suggestions even more helpful. That means referencing Gmail

[02:00:41] That means referencing Gmail receipts, Calendar and much more

[02:00:41] receipts, Calendar and much more to personalize.

[02:00:43] to personalize. And boom, there it is.

[02:00:43] And boom, there it is. It looks like it's ready.

[02:00:48] It looks like it's ready. All right. It's got a beautiful

[02:00:49] All right. It's got a beautiful planner, already taken into

[02:00:50] planner, already taken into account driving times and

[02:00:52] account driving times and weather.

[02:00:52] weather. And Search knows that I have two

[02:00:54] And Search knows that I have two kids. And that they love

[02:00:55] kids. And that they love animals and that my oldest is

[02:00:56] animals and that my oldest is learning to play chess. So

[02:00:58] learning to play chess. So actually this second one here is

[02:00:59] actually this second one here is kind of awesome for my oldest so

[02:01:01] kind of awesome for my oldest so I'm going to go ahead and heart

[02:01:04] I'm going to go ahead and heart that. But, you know, two kids,

[02:01:05] that. But, you know, two kids, got to make them both happy so

[02:01:07] got to make them both happy so I'm going to lock in the Happy

[02:01:10] I'm going to lock in the Happy Hollow Park & Zoo right there.

[02:01:12] Hollow Park & Zoo right there. And because this syncs with my

[02:01:13] And because this syncs with my calendar, you can see it's

[02:01:15] calendar, you can see it's already blocked off my afternoon

[02:01:16] already blocked off my afternoon to meet a friend to watch this

[02:01:18] to meet a friend to watch this game down here.

[02:01:19] game down here. And below it's got all of these

[02:01:21] And below it's got all of these cool restaurant reservations and

[02:01:22] cool restaurant reservations and beautifully laid out on Maps.

[02:01:25] beautifully laid out on Maps. Now having seen this agents, I

[02:01:26] Now having seen this agents, I kind of want the Madam President

[02:01:27] kind of want the Madam President to be higher up, and also, my

[02:01:30] to be higher up, and also, my wife, and I try to do Friday

[02:01:31] wife, and I try to do Friday date nights. So I'm going to

[02:01:32] date nights. So I'm going to keep customizing this and just

[02:01:35] keep customizing this and just for the sake of speed, I'm going

[02:01:36] for the sake of speed, I'm going to paste in here one more

[02:01:38] to paste in here one more question to add Friday night

[02:01:40] question to add Friday night date each week and move it to

[02:01:41] date each week and move it to the top.

[02:01:42] the top. And just like before, it's

[02:01:44] And just like before, it's thinking through what's needed

[02:01:45] thinking through what's needed to adjust the Planner, looking

[02:01:46] to adjust the Planner, looking up realtime information, even

[02:01:48] up realtime information, even checking my preferences again.

[02:01:49] checking my preferences again. Wow. It was so fast.

[02:01:51] Wow. It was so fast. So it's able to use all kinds of

[02:01:53] So it's able to use all kinds of information from Google to build

[02:01:54] information from Google to build this on the fly and now, you can

[02:01:56] this on the fly and now, you can see the map right at the top,

[02:01:59] see the map right at the top, and Friday date night tab, right

[02:02:01] and Friday date night tab, right here.

[02:02:02] here. So I can scroll down, I can see

[02:02:04] So I can scroll down, I can see all of these really great

[02:02:05] all of these really great places, after the baby-sitter

[02:02:07] places, after the baby-sitter arrives, here's some awesome

[02:02:08] arrives, here's some awesome restaurants. I'll go ahead and

[02:02:09] restaurants. I'll go ahead and select it, and we're ready to

[02:02:12] select it, and we're ready to go.

[02:02:12] go. Now, a weekend plan is not

[02:02:14] Now, a weekend plan is not complete until I get my wife

[02:02:16] complete until I get my wife Danielle's seal of approval so I

[02:02:17] Danielle's seal of approval so I can share this app with her,

[02:02:20] can share this app with her, which I'm going to do right

[02:02:22] which I'm going to do right here. And copy it. Okay. And

[02:02:24] here. And copy it. Okay. And now, I'm actually going to pull

[02:02:25] now, I'm actually going to pull it up on a phone. When I send

[02:02:26] it up on a phone. When I send this to her this is exactly what

[02:02:28] this to her this is exactly what she's going to see on her phone

[02:02:37] she's going to see on her phone All right. So Danielle is in,

[02:02:38] All right. So Danielle is in, but it also looks like she might

[02:02:39] but it also looks like she might have some feedback for me when I

[02:02:41] have some feedback for me when I get home, but we're going to

[02:02:43] get home, but we're going to deal with that kind of a little

[02:02:44] deal with that kind of a little bit later. So all I have to do

[02:02:47] bit later. So all I have to do is add this to my Calendar.

[02:02:48] is add this to my Calendar. Search will add it to all of our

[02:02:50] Search will add it to all of our family calendars, and we're

[02:02:51] family calendars, and we're ready to go and next weekend

[02:02:53] ready to go and next weekend I'll be able to come back and

[02:02:54] I'll be able to come back and plan a whole new weekend with my

[02:02:55] plan a whole new weekend with my family just like this.

[02:02:57] family just like this. So we're bringing Antigravity

[02:02:58] So we're bringing Antigravity into Search, so you'll get

[02:02:59] into Search, so you'll get generative UI this summer.

[02:03:01] generative UI this summer. Starting with subscribers,

[02:03:03] Starting with subscribers, you'll be able to custom-build

[02:03:05] you'll be able to custom-build experiences just like this in

[02:03:06] experiences just like this in the coming months.

[02:03:07] the coming months. [Applause]

[02:03:16] [Applause] From Search agents to agentic

[02:03:18] From Search agents to agentic coding, this is an AI Search

[02:03:20] coding, this is an AI Search that does more for you, whatever

[02:03:22] that does more for you, whatever your question.

[02:03:25] your question. Agentic capabilities are going

[02:03:25] Agentic capabilities are going to transform all the ways you

[02:03:26] to transform all the ways you use Search, including how you

[02:03:27] use Search, including how you shop.

[02:03:27] shop. To tell you more, here's Vidhya

[02:03:29] To tell you more, here's Vidhya ♪

[02:03:30] ♪ ♪

[02:03:30] ♪ [Applause]

[02:03:43] [Applause] >> VIDHYA SRINIVASAN: With

[02:03:45] >> VIDHYA SRINIVASAN: With incredible advancements in AI,

[02:03:47] incredible advancements in AI, we are entering a whole new era.

[02:03:49] we are entering a whole new era. We've been building the

[02:03:51] We've been building the foundation for agentic commerce,

[02:03:55] foundation for agentic commerce, and now, we're bringing that

[02:03:56] and now, we're bringing that future right to you. People

[02:03:57] future right to you. People shop across Google over a

[02:03:59] shop across Google over a billion times a day, and for

[02:04:01] billion times a day, and for years, we've connected you with

[02:04:03] years, we've connected you with brands and retailers to help you

[02:04:05] brands and retailers to help you get exactly what you're looking

[02:04:07] get exactly what you're looking for. It starts with our

[02:04:09] for. It starts with our Shopping Graph, the world's most

[02:04:11] Shopping Graph, the world's most comprehensive catalog of

[02:04:15] comprehensive catalog of products. It now has over 60

[02:04:17] products. It now has over 60 billion listings and they are

[02:04:19] billion listings and they are constantly updated. Combine

[02:04:20] constantly updated. Combine that, the scale of the Shopping

[02:04:21] that, the scale of the Shopping Graph, with our most advanced

[02:04:22] Graph, with our most advanced Gemini models and you get

[02:04:25] Gemini models and you get entirely new ways to shop. More

[02:04:27] entirely new ways to shop. More powerful, more intelligent, and

[02:04:31] powerful, more intelligent, and more fun! Now, when it comes to

[02:04:32] more fun! Now, when it comes to agentic commerce, we are focused

[02:04:34] agentic commerce, we are focused on delivering three key building

[02:04:35] on delivering three key building blocks to make this future a

[02:04:39] blocks to make this future a reality.

[02:04:39] reality. First, the Universal Commerce

[02:04:43] First, the Universal Commerce Protocol, or UCP for short. UCP

[02:04:47] Protocol, or UCP for short. UCP does for agentic commerce what

[02:04:51] does for agentic commerce what HTTP did for the web: It gives

[02:04:52] HTTP did for the web: It gives agents and systems a common

[02:04:54] agents and systems a common language. It's an open-source

[02:04:55] language. It's an open-source standard that allows all of the

[02:04:58] standard that allows all of the key players to work across the

[02:05:01] key players to work across the entire shopping journey. That

[02:05:04] entire shopping journey. That means it makes everything from

[02:05:05] means it makes everything from product research to checkout to

[02:05:06] product research to checkout to shipment tracking totally

[02:05:07] shipment tracking totally seamless.

[02:05:09] seamless. And we've been thrilled to see

[02:05:11] And we've been thrilled to see the entire industry rally behind

[02:05:15] the entire industry rally behind it.

[02:05:15] it. We co-developed and launched UCP

[02:05:16] We co-developed and launched UCP with an incredible group of

[02:05:18] with an incredible group of founding partners, and recently

[02:05:23] founding partners, and recently welcomed Amazon, Meta,

[02:05:26] welcomed Amazon, Meta, Microsoft, Salesforce, and

[02:05:30] Microsoft, Salesforce, and Stripe.

[02:05:35] Stripe. [Applause]

[02:05:39] [Applause] They're going to help continue

[02:05:40] They're going to help continue steering this open standard.

[02:05:41] steering this open standard. It may just be the first time

[02:05:42] It may just be the first time we've all agreed on something!

[02:05:46] we've all agreed on something! And now, we are bringing UCP to

[02:05:48] And now, we are bringing UCP to even more verticals, like

[02:05:54] even more verticals, like hotels and local food delivery

[02:05:55] hotels and local food delivery providers and to YouTube and

[02:05:56] providers and to YouTube and more products. We are also

[02:05:57] more products. We are also expanding UCP-powered

[02:05:59] expanding UCP-powered experiences on Google products

[02:06:01] experiences on Google products to more regions, like Canada,

[02:06:03] to more regions, like Canada, Australia, and the U.K. in the

[02:06:04] Australia, and the U.K. in the coming months.

[02:06:08] coming months. This brings us to our second

[02:06:13] This brings us to our second building block: The agent

[02:06:14] building block: The agent payments protocol or AP2.

[02:06:17] payments protocol or AP2. So when it comes to agentic

[02:06:18] So when it comes to agentic payments, the number one

[02:06:20] payments, the number one question we hear is, "how do I

[02:06:24] question we hear is, "how do I know it won't just go off and

[02:06:25] know it won't just go off and buy something I don't want?"

[02:06:28] buy something I don't want?" Motorcycle anyone?

[02:06:30] Motorcycle anyone? It's a fair question, and it's

[02:06:31] It's a fair question, and it's why we created AP2. It's

[02:06:34] why we created AP2. It's designed so your AI agent can

[02:06:35] designed so your AI agent can securely make payments on your

[02:06:39] securely make payments on your behalf, but always under your

[02:06:41] behalf, but always under your control.

[02:06:42] control. For us, it comes down to two

[02:06:44] For us, it comes down to two things: Setting boundaries and

[02:06:47] things: Setting boundaries and ensuring accountability. First,

[02:06:49] ensuring accountability. First, let's talk about boundaries. We

[02:06:51] let's talk about boundaries. We are now making it easy to set

[02:06:53] are now making it easy to set strict guardrails. Just tell

[02:06:54] strict guardrails. Just tell your agent the specific brands

[02:06:56] your agent the specific brands and products you want and how

[02:07:01] and products you want and how much you want to spend. It

[02:07:02] much you want to spend. It automatically makes the purchase

[02:07:05] automatically makes the purchase if the criteria are met. But

[02:07:07] if the criteria are met. But boundaries only work if there's

[02:07:09] boundaries only work if there's accountability. AP2 creates a

[02:07:11] accountability. AP2 creates a transparent, verifiable link

[02:07:13] transparent, verifiable link between you, the merchant and

[02:07:18] between you, the merchant and the payment processors. It uses

[02:07:19] the payment processors. It uses privacy-preserving technology to

[02:07:19] privacy-preserving technology to keep your data safe and with

[02:07:21] keep your data safe and with tamper-proof digital mandates,

[02:07:24] tamper-proof digital mandates, AP2 ensures the agent is always

[02:07:27] AP2 ensures the agent is always acting on your behalf and giving

[02:07:28] acting on your behalf and giving you a permanent digital paper

[02:07:30] you a permanent digital paper trail.

[02:07:33] trail. So if you ever need to make a

[02:07:33] So if you ever need to make a return, you and the merchant are

[02:07:35] return, you and the merchant are looking at the same record.

[02:07:38] looking at the same record. Your payment info stays shielded

[02:07:39] Your payment info stays shielded and your data stays private and

[02:07:42] and your data stays private and your purchases stay secure.

[02:07:46] your purchases stay secure. We'll begin bringing AP2 to

[02:07:47] We'll begin bringing AP2 to Google products in the coming

[02:07:51] Google products in the coming months, starting with Gemini

[02:07:53] months, starting with Gemini Spark.

[02:07:55] Spark. [Applause]

[02:07:58] [Applause] Now, these protocols are

[02:07:58] Now, these protocols are powering the foundation of this

[02:08:01] powering the foundation of this new era, which brings us to our

[02:08:03] new era, which brings us to our final building block of agentic

[02:08:05] final building block of agentic commerce. I'm excited to

[02:08:08] commerce. I'm excited to announce the Universal Cart, a

[02:08:09] announce the Universal Cart, a truly intelligent shopping cart.

[02:08:14] truly intelligent shopping cart. It works across merchants and

[02:08:18] It works across merchants and across services. You will be

[02:08:19] across services. You will be able to add things to your cart

[02:08:20] able to add things to your cart while you're browsing Search,

[02:08:23] while you're browsing Search, chatting with Gemini, watching

[02:08:24] chatting with Gemini, watching YouTube or even reading your

[02:08:27] YouTube or even reading your Gmail. The moment you add a

[02:08:28] Gmail. The moment you add a product, your cart goes to work

[02:08:33] product, your cart goes to work for you in the background. It

[02:08:34] for you in the background. It finds deals and price drops, it

[02:08:36] finds deals and price drops, it gives you insights on price

[02:08:38] gives you insights on price history, and alerts you when

[02:08:39] history, and alerts you when something comes back in stock.

[02:08:43] something comes back in stock. It all runs on our Gemini models

[02:08:47] It all runs on our Gemini models so your cart gets even smarter

[02:08:48] so your cart gets even smarter as the models improve.

[02:08:51] as the models improve. So just think of it as shopping

[02:08:53] So just think of it as shopping with superpowers. Another game

[02:08:54] with superpowers. Another game changer is how it applies

[02:08:58] changer is how it applies intelligent reasoning. Let's

[02:09:00] intelligent reasoning. Let's say you're building your first

[02:09:00] say you're building your first custom PC. You see a

[02:09:01] custom PC. You see a motherboard with great reviews

[02:09:03] motherboard with great reviews and add it to your cart. You

[02:09:05] and add it to your cart. You had already picked out a

[02:09:06] had already picked out a processor, but you didn't

[02:09:08] processor, but you didn't realize is that the processor

[02:09:09] realize is that the processor needs a motherboard with a

[02:09:11] needs a motherboard with a different type of socket. Your

[02:09:14] different type of socket. Your cart catches this for you and

[02:09:14] cart catches this for you and suggests an alternative,

[02:09:15] suggests an alternative, preventing a problem you didn't

[02:09:23] preventing a problem you didn't really see coming! Next, my

[02:09:24] really see coming! Next, my favorite feature is how the cart

[02:09:24] favorite feature is how the cart can actually find hidden savings

[02:09:25] can actually find hidden savings for you.

[02:09:29] for you. If you're like me, you have

[02:09:30] If you're like me, you have different payment cards with

[02:09:31] different payment cards with different perks that are hard to

[02:09:35] different perks that are hard to keep track of. Now, you don't

[02:09:39] keep track of. Now, you don't have to. The cart can do it for

[02:09:39] have to. The cart can do it for you because it's built on Google

[02:09:41] you because it's built on Google wallet.

[02:09:41] wallet. Here's an example. Right here,

[02:09:44] Here's an example. Right here, there's an offer from Target for

[02:09:46] there's an offer from Target for some products that were added to

[02:09:48] some products that were added to the cart earlier in the week. I

[02:09:49] the cart earlier in the week. I see it so I want to get them

[02:09:52] see it so I want to get them now. And UCP makes checkout

[02:09:57] now. And UCP makes checkout from your cart super smooth.

[02:09:57] from your cart super smooth. For many of your favorite

[02:09:59] For many of your favorite brands, you can check out right

[02:10:01] brands, you can check out right on Google in just a few taps

[02:10:01] on Google in just a few taps with Google Pay, or you can

[02:10:03] with Google Pay, or you can transfer the items straight to

[02:10:04] transfer the items straight to the retailer's site and buy them

[02:10:07] the retailer's site and buy them there.

[02:10:09] there. I'm excited to announce that we

[02:10:11] I'm excited to announce that we are rolling out the Universal

[02:10:12] are rolling out the Universal Cart in the U.S. across Search

[02:10:15] Cart in the U.S. across Search and the Gemini app this summer

[02:10:16] and the Gemini app this summer with YouTube and Gmail to

[02:10:18] with YouTube and Gmail to follow.

[02:10:21] follow. [Applause]

[02:10:27] [Applause] These are the building blocks

[02:10:29] These are the building blocks that we showed you today, and

[02:10:30] that we showed you today, and they don't just create the

[02:10:30] they don't just create the foundation for agentic commerce,

[02:10:32] foundation for agentic commerce, they'll transform how you shop

[02:10:35] they'll transform how you shop on Google, making it more

[02:10:37] on Google, making it more powerful, more intelligent and a

[02:10:38] powerful, more intelligent and a lot more fun.

[02:10:41] lot more fun. Happy shopping and back to you,

[02:10:43] Happy shopping and back to you, Liz!

[02:10:43] Liz! [Applause]

[02:10:50] [Applause] >> LIZ REID: Today, you've seen

[02:10:51] >> LIZ REID: Today, you've seen how we're bringing together the

[02:10:52] how we're bringing together the best of a search engine with the

[02:10:54] best of a search engine with the best of AI to build a Google

[02:10:56] best of AI to build a Google Search that's more helpful and

[02:10:58] Search that's more helpful and powerful than ever before, where

[02:11:00] powerful than ever before, where Search agents work for you

[02:11:01] Search agents work for you around the clock, where agentic

[02:11:05] around the clock, where agentic coding means that Search can

[02:11:06] coding means that Search can build experiences as unique as

[02:11:09] build experiences as unique as your questions, and where the

[02:11:10] your questions, and where the power of agentic commerce isn't

[02:11:13] power of agentic commerce isn't in the distant future; it's here

[02:11:15] in the distant future; it's here now.

[02:11:17] now. So in this chapter of AI Search,

[02:11:20] So in this chapter of AI Search, Google can help you find,

[02:11:23] Google can help you find, understand, build, do anything.

[02:11:25] understand, build, do anything. So go ahead; just ask.

[02:11:26] So go ahead; just ask. ♪

[02:11:26] ♪ ♪

[02:11:35] ♪ [Applause]

[02:11:35] [Applause] ♪

[02:11:35] ♪ ♪

[02:12:53] ♪ [Applause]

[02:12:59] [Applause] >> JOSH WOODWARD: All right! Hi,

[02:13:00] >> JOSH WOODWARD: All right! Hi, everyone! Great to see you

[02:13:02] everyone! Great to see you again. And you've already seen

[02:13:04] again. And you've already seen some amazing breakthroughs this

[02:13:05] some amazing breakthroughs this morning, and we're just getting

[02:13:07] morning, and we're just getting started.

[02:13:07] started. We have a lot more to show you

[02:13:08] We have a lot more to show you in the Gemini app so let's jump

[02:13:11] in the Gemini app so let's jump right in. Over the last summer,

[02:13:15] right in. Over the last summer, Gemini has had incredible

[02:13:17] Gemini has had incredible momentum.

[02:13:17] momentum. More than 900 million users are

[02:13:19] More than 900 million users are coming to the Gemini app every

[02:13:22] coming to the Gemini app every month, and a big part of that

[02:13:22] month, and a big part of that growth is how fast we're

[02:13:26] growth is how fast we're shipping.

[02:13:26] shipping. A year ago, none of these

[02:13:28] A year ago, none of these features existed. And now,

[02:13:30] features existed. And now, Gemini has become the ultimate

[02:13:33] Gemini has become the ultimate creative tool for everyone. You

[02:13:35] creative tool for everyone. You can create images, videos and

[02:13:36] can create images, videos and music in ways people never

[02:13:39] music in ways people never thought were possible. We made

[02:13:43] thought were possible. We made Gemini even more tailored with

[02:13:44] Gemini even more tailored with Personal Intelligence, allowing

[02:13:45] Personal Intelligence, allowing you to securely connect your

[02:13:49] you to securely connect your Gmail, Photos, and other apps so

[02:13:49] Gmail, Photos, and other apps so you can get customized help.

[02:13:52] you can get customized help. Just last week, we expanded

[02:13:55] Just last week, we expanded Personal Intelligence to users

[02:13:59] Personal Intelligence to users Around the world directly in the

[02:14:03] Around the world directly in the app.

[02:14:03] app. Millions of people are using it

[02:14:04] Millions of people are using it every single day.

[02:14:04] every single day. They found it so helpful for

[02:14:06] They found it so helpful for things like personalized product

[02:14:07] things like personalized product and trip recommendations or

[02:14:11] and trip recommendations or Acting as a thought partner for

[02:14:12] Acting as a thought partner for navigating big decisions in life

[02:14:13] navigating big decisions in life like a career change or in my

[02:14:15] like a career change or in my case, finding the right set of

[02:14:16] case, finding the right set of tires for our super cool min van

[02:14:20] tires for our super cool min van that's going to be at that block

[02:14:21] that's going to be at that block party. We've made Gemini superb

[02:14:24] party. We've made Gemini superb for students also over the last

[02:14:25] for students also over the last year. You can use it to create

[02:14:27] year. You can use it to create guided learning, generate

[02:14:27] guided learning, generate practice tests, and even use

[02:14:29] practice tests, and even use dedicated notebooks to keep all

[02:14:31] dedicated notebooks to keep all your notes and assignments in

[02:14:33] your notes and assignments in one place. And these same

[02:14:35] one place. And these same notebooks sync directly to

[02:14:38] notebooks sync directly to NotebookLM. Just on its own,

[02:14:40] NotebookLM. Just on its own, NotebookLM has now been used to

[02:14:41] NotebookLM has now been used to create more than 1.5 billion

[02:14:45] create more than 1.5 billion notebooks, podcasts, slide decks

[02:14:47] notebooks, podcasts, slide decks and more, turning complex

[02:14:51] and more, turning complex information into easy to

[02:14:52] information into easy to understand knowledge. And as

[02:14:53] understand knowledge. And as you heard from Sundar earlier,

[02:14:55] you heard from Sundar earlier, we're rolling out our new Gemini

[02:14:58] we're rolling out our new Gemini 3.5 Flash model, which is

[02:14:59] 3.5 Flash model, which is unlocking a new set of features

[02:15:01] unlocking a new set of features that I'm going to show you

[02:15:04] that I'm going to show you today. All of this is happening

[02:15:05] today. All of this is happening in a Gemini app that's now

[02:15:07] in a Gemini app that's now available in more than 230

[02:15:10] available in more than 230 countries and over 70 languages,

[02:15:14] countries and over 70 languages, making Gemini the most widely

[02:15:15] making Gemini the most widely available AI system in the

[02:15:18] available AI system in the world.

[02:15:19] world. [Applause]

[02:15:23] [Applause] Gemini is becoming that

[02:15:25] Gemini is becoming that universal assistant people are

[02:15:26] universal assistant people are turning to. It's designed for

[02:15:28] turning to. It's designed for their everyday lives, and

[02:15:29] their everyday lives, and speaking of design, that brings

[02:15:30] speaking of design, that brings us to the first of our three big

[02:15:34] us to the first of our three big updates today. Today, I'm

[02:15:35] updates today. Today, I'm excited to announce that we've

[02:15:37] excited to announce that we've completely redesigned the Gemini

[02:15:39] completely redesigned the Gemini experience from the ground up.

[02:15:41] experience from the ground up. From the moment you open it,

[02:15:45] From the moment you open it, we've greeted you with a

[02:15:47] we've greeted you with a stunning new design language we

[02:15:48] stunning new design language we call Neural Expressive. We've

[02:15:49] call Neural Expressive. We've added fluid animations, vibrant

[02:15:52] added fluid animations, vibrant colors, new topography, haptic

[02:15:55] colors, new topography, haptic feedback throughout the app.

[02:15:56] feedback throughout the app. But we all know that good design

[02:15:58] But we all know that good design isn't just about how it looks;

[02:16:00] isn't just about how it looks; good design is about how it

[02:16:03] good design is about how it works. So we've involved the

[02:16:05] works. So we've involved the entire experience. We've made

[02:16:07] entire experience. We've made it easier to discover and

[02:16:09] it easier to discover and generate those gorgeous images,

[02:16:11] generate those gorgeous images, videos and music with built-in

[02:16:14] videos and music with built-in templates that you can easily

[02:16:16] templates that you can easily remix. And we've completely

[02:16:17] remix. And we've completely transformed the Gemini Live

[02:16:18] transformed the Gemini Live experience. It now opens up

[02:16:20] experience. It now opens up immediately and in line. And

[02:16:22] immediately and in line. And soon, you'll be able to pick a

[02:16:24] soon, you'll be able to pick a regional dialect that resonates

[02:16:26] regional dialect that resonates with you.

[02:16:28] with you. >> You've got a right good mix

[02:16:29] >> You've got a right good mix of different accents knocking

[02:16:31] of different accents knocking about like this one from

[02:16:33] about like this one from Liverpool.

[02:16:34] Liverpool. [Speaking various languages]

[02:16:51] [Speaking various languages] >> JOSH WOODWARD: All right.

[02:16:52] >> JOSH WOODWARD: All right. Pretty cool.

[02:16:52] Pretty cool. [Applause]

[02:16:55] [Applause] We'll be rolling out these and

[02:16:58] We'll be rolling out these and many more regional dialects in

[02:16:59] many more regional dialects in the coming weeks, and my

[02:17:02] the coming weeks, and my favorite part is how we handle

[02:17:04] favorite part is how we handle model responses with our new

[02:17:07] model responses with our new Neural Expressive design

[02:17:08] Neural Expressive design language. That's where it

[02:17:08] language. That's where it really comes to life.

[02:17:09] really comes to life. You won't see a wall of text

[02:17:12] You won't see a wall of text anymore. Instead, Gemini will

[02:17:13] anymore. Instead, Gemini will carefully lay out its response

[02:17:16] carefully lay out its response in real time, just for you, like

[02:17:20] in real time, just for you, like the generative UI you saw in

[02:17:25] the generative UI you saw in Search.

[02:17:27] Search. As you scroll, you might see

[02:17:27] As you scroll, you might see interactive images that are

[02:17:28] interactive images that are custom generated by Gemini. You

[02:17:29] custom generated by Gemini. You can drill into them and explore

[02:17:31] can drill into them and explore information on an entirely new

[02:17:32] information on an entirely new level. You might even see

[02:17:33] level. You might even see timelines that you can quickly

[02:17:35] timelines that you can quickly skim, or embedded videos. So

[02:17:37] skim, or embedded videos. So whether you're in dark mode or

[02:17:38] whether you're in dark mode or light mode, the entire

[02:17:40] light mode, the entire experience feels fluid,

[02:17:43] experience feels fluid, futuristic, and incredibly

[02:17:46] futuristic, and incredibly natural. Best of all? Neural

[02:17:47] natural. Best of all? Neural Expressive is rolling out

[02:17:50] Expressive is rolling out globally on Android, iOS, and

[02:17:52] globally on Android, iOS, and the Web starting right now.

[02:17:52] the Web starting right now. [Applause]

[02:17:59] [Applause] With a completely redesigned

[02:18:01] With a completely redesigned Gemini app as the new base, we

[02:18:05] Gemini app as the new base, we can go even further with

[02:18:05] can go even further with Gemini's creative capabilities.

[02:18:06] Gemini's creative capabilities. And that's update number 2.

[02:18:10] And that's update number 2. Gemini Omni is coming to the

[02:18:11] Gemini Omni is coming to the Gemini app for paid subscribers

[02:18:13] Gemini app for paid subscribers today. It's going to let you

[02:18:15] today. It's going to let you bring your ideas to life using

[02:18:17] bring your ideas to life using any combination of text, images

[02:18:19] any combination of text, images and video inputs. As I've been

[02:18:22] and video inputs. As I've been using it, it feels like the Nano

[02:18:24] using it, it feels like the Nano Banana for video moment is here.

[02:18:26] Banana for video moment is here. It's never been so easy to

[02:18:29] It's never been so easy to create and edit videos. Let's

[02:18:30] create and edit videos. Let's look at how this plays out in

[02:18:32] look at how this plays out in the real world. I want you to

[02:18:34] the real world. I want you to meet Sashu. She's working on a

[02:18:35] meet Sashu. She's working on a new song and wants to create a

[02:18:38] new song and wants to create a quick video teaser. She shares

[02:18:45] quick video teaser. She shares this raw video, she adds some

[02:18:45] this raw video, she adds some reference visuals and this is

[02:18:46] reference visuals and this is the coolest part. She can

[02:18:47] the coolest part. She can transform the style of the

[02:18:52] transform the style of the video, even switch the camera

[02:18:55] video, even switch the camera angle to a 360-degree shot. And

[02:18:56] angle to a 360-degree shot. And Gemini puts it all together.

[02:18:57] Gemini puts it all together. Let's take a look at what it

[02:18:58] Let's take a look at what it looks like.

[02:18:58] looks like. ♪

[02:18:58] ♪ ♪

[02:19:11] ♪ [Applause]

[02:19:11] [Applause] >> JOSH WOODWARD: All done

[02:19:12] >> JOSH WOODWARD: All done within Gemini.

[02:19:14] within Gemini. As you can see, Omni understands

[02:19:16] As you can see, Omni understands the physics of her movement,

[02:19:17] the physics of her movement, intelligently layering those

[02:19:18] intelligently layering those effects into the real world

[02:19:20] effects into the real world without losing the soul of the

[02:19:23] without losing the soul of the shot. Creating, remixing, and

[02:19:26] shot. Creating, remixing, and Editing a video has never been

[02:19:28] Editing a video has never been easier. So whatever vision is

[02:19:29] easier. So whatever vision is in her head, she can now use

[02:19:31] in her head, she can now use Gemini to make it real.

[02:19:37] Gemini to make it real. Google AI Plus, Pro and Ultra

[02:19:37] Google AI Plus, Pro and Ultra subscribers around the worl

[02:19:41] subscribers around the worl Can try Gemini Omni today right

[02:19:42] Can try Gemini Omni today right in the app.

[02:19:43] in the app. [Applause]

[02:19:50] [Applause] The third update today is about

[02:19:53] The third update today is about how agents are coming to Gemini.

[02:19:53] how agents are coming to Gemini. This is a big shift for Gemini,

[02:19:54] This is a big shift for Gemini, because agents don't just answer

[02:19:55] because agents don't just answer questions; they proactively work

[02:19:58] questions; they proactively work on your behalf.

[02:19:59] on your behalf. And to show you what that means,

[02:20:01] And to show you what that means, I want to introduce one of our

[02:20:02] I want to introduce one of our newest out of the box agents

[02:20:03] newest out of the box agents called the daily brief.

[02:20:05] called the daily brief. It's a personalized digest

[02:20:06] It's a personalized digest that's designed to be your first

[02:20:08] that's designed to be your first stop every morning. Here's how

[02:20:12] stop every morning. Here's how it works. You can see here it's

[02:20:14] it works. You can see here it's synthesizing information from

[02:20:15] synthesizing information from across my inbox, my calendar, my

[02:20:18] across my inbox, my calendar, my tasks, it's finding the most

[02:20:19] tasks, it's finding the most important things for me to be

[02:20:23] important things for me to be aware of.

[02:20:23] aware of. And I could totally forget these

[02:20:24] And I could totally forget these things like returning that back.

[02:20:25] things like returning that back. It's not just summarizing the

[02:20:27] It's not just summarizing the data, though; it's organizing it

[02:20:29] data, though; it's organizing it by topic. It's even suggesting

[02:20:31] by topic. It's even suggesting the next steps. And with this

[02:20:33] the next steps. And with this travel info, I can just take the

[02:20:35] travel info, I can just take the next step right in line. All of

[02:20:38] next step right in line. All of it is super concise in this

[02:20:41] it is super concise in this morning digest that's built for

[02:20:43] morning digest that's built for skimming.

[02:20:43] skimming. And I can't wait for you to try

[02:20:44] And I can't wait for you to try it out.

[02:20:45] it out. The Daily Brief is rolling out

[02:20:48] The Daily Brief is rolling out today also to

[02:20:52] today also to Google AI Plus, Pro, and Ultra

[02:20:53] Google AI Plus, Pro, and Ultra subscribers starting right here

[02:20:53] subscribers starting right here in the U.S.

[02:20:54] in the U.S. And that's just the help you're

[02:20:55] And that's just the help you're going to get before breakfast.

[02:20:58] going to get before breakfast. Beyond the Daily Brief, we're

[02:20:59] Beyond the Daily Brief, we're allowing power users to create

[02:21:01] allowing power users to create their own custom workflows with

[02:21:05] their own custom workflows with Gemini Spark like I showed you

[02:21:09] Gemini Spark like I showed you earlier.

[02:21:09] earlier. You'll remember at the beginning

[02:21:11] You'll remember at the beginning Of the show I sent a few tasks

[02:21:14] Of the show I sent a few tasks off so let's go check in on

[02:21:17] off so let's go check in on them, let's see how they've

[02:21:19] them, let's see how they've completed. And you'll remember

[02:21:20] completed. And you'll remember when I did this earlier, I did

[02:21:21] when I did this earlier, I did it from the phone so we're going

[02:21:23] it from the phone so we're going to actually pull it up on the

[02:21:25] to actually pull it up on the laptop to see how it thinks

[02:21:27] laptop to see how it thinks across.

[02:21:27] across. And if I open it up here on the

[02:21:29] And if I open it up here on the laptop you can see it's

[02:21:29] laptop you can see it's subdivided those three tasks

[02:21:30] subdivided those three tasks right here, one, two, three, and

[02:21:32] right here, one, two, three, and it's actually got a really nice

[02:21:34] it's actually got a really nice feature. There's certain tasks

[02:21:36] feature. There's certain tasks it will actually ask for your

[02:21:37] it will actually ask for your input so you can approve it, so

[02:21:39] input so you can approve it, so it doesn't just go off and do

[02:21:40] it doesn't just go off and do things you don't want.

[02:21:41] things you don't want. But I'll check on this one here

[02:21:43] But I'll check on this one here This is our school year planning

[02:21:46] This is our school year planning checklist and I asked it

[02:21:47] checklist and I asked it remember to create a docket of

[02:21:47] remember to create a docket of all the things I didn't want to

[02:21:50] all the things I didn't want to forget between now and the end

[02:21:51] forget between now and the end of the year.

[02:21:52] of the year. And so I'll open this one up,

[02:21:54] And so I'll open this one up, and what's amazing about this is

[02:21:55] and what's amazing about this is it takes advantage of all the

[02:21:56] it takes advantage of all the Google Docs formatting so I can

[02:21:57] Google Docs formatting so I can immediately click in and see the

[02:21:59] immediately click in and see the checklist here for all our

[02:22:00] checklist here for all our various kids, all subdivided

[02:22:01] various kids, all subdivided like this, and easily go one by

[02:22:02] like this, and easily go one by one, the date, the activity, the

[02:22:04] one, the date, the activity, the color coding, all integrated in

[02:22:05] color coding, all integrated in one spot.

[02:22:08] one spot. Pretty incredible how much time

[02:22:09] Pretty incredible how much time this can save.

[02:22:11] this can save. [Applause]

[02:22:11] [Applause] Now, what comes across all of

[02:22:13] Now, what comes across all of these cases is that we want to

[02:22:16] these cases is that we want to make agents easy to use and safe

[02:22:19] make agents easy to use and safe and secure. As a reminder,

[02:22:20] and secure. As a reminder, Spark is going to start rolling

[02:22:24] Spark is going to start rolling out to trusted testers this

[02:22:25] out to trusted testers this week, and we're planning to roll

[02:22:27] week, and we're planning to roll it out as a beta to Google AI

[02:22:32] it out as a beta to Google AI Ultra subscribers next week in

[02:22:34] Ultra subscribers next week in the U.S., and we'll be bringing

[02:22:35] the U.S., and we'll be bringing a version of Spark also to

[02:22:37] a version of Spark also to Gemini Workspace, as well as

[02:22:38] Gemini Workspace, as well as Gemini Enterprise.

[02:22:39] Gemini Enterprise. And this is just the beginning.

[02:22:39] And this is just the beginning. We've got a packed road map of

[02:22:41] We've got a packed road map of features we'll be shipping

[02:22:42] features we'll be shipping throughout the summer. It's

[02:22:43] throughout the summer. It's going to be amazing. I'm really

[02:22:45] going to be amazing. I'm really excited about the MCP

[02:22:47] excited about the MCP integration. That's going to

[02:22:48] integration. That's going to enable Spark to handle so many

[02:22:50] enable Spark to handle so many tasks in ways that are even more

[02:22:54] tasks in ways that are even more proactive and powerful.

[02:22:56] proactive and powerful. Imagine, Spark will be able to

[02:22:56] Imagine, Spark will be able to look ahead on your calendar and

[02:22:58] look ahead on your calendar and see that you're on snack duty,

[02:23:00] see that you're on snack duty, or for your kids' T-ball game on

[02:23:03] or for your kids' T-ball game on Saturday.

[02:23:04] Saturday. They can go ahead, and

[02:23:05] They can go ahead, and proactively set up your

[02:23:06] proactively set up your instacart order on its own so

[02:23:07] instacart order on its own so you don't forget those snacks.

[02:23:09] you don't forget those snacks. It will even remember to pick

[02:23:10] It will even remember to pick snacks that don't have nuts in

[02:23:15] snacks that don't have nuts in them.

[02:23:16] them. And we have this incredible

[02:23:16] And we have this incredible lineup of partners that are

[02:23:17] lineup of partners that are coming that will be integrating

[02:23:18] coming that will be integrating with Spark over the summer.

[02:23:21] with Spark over the summer. Now, I know I promised just

[02:23:23] Now, I know I promised just three updates, but we have one

[02:23:25] three updates, but we have one more. Last month, we dropped

[02:23:26] more. Last month, we dropped the Gemini app for Mac OS. Here

[02:23:28] the Gemini app for Mac OS. Here it is on the screen. It's

[02:23:30] it is on the screen. It's gorgeous. This is a small team

[02:23:31] gorgeous. This is a small team that built this native app from

[02:23:35] that built this native app from scratch, using Antigravity.

[02:23:36] scratch, using Antigravity. They did over 100 features in

[02:23:38] They did over 100 features in less than 100 days. Now, two

[02:23:42] less than 100 days. Now, two weekends ago we were hacking on

[02:23:43] weekends ago we were hacking on the Mac app and we came up with

[02:23:45] the Mac app and we came up with something cool and we wanted to

[02:23:46] something cool and we wanted to sneak it into the show. Do you

[02:23:47] sneak it into the show. Do you all want to see it live?

[02:23:49] all want to see it live? All right.

[02:23:51] All right. [Applause]

[02:23:51] [Applause] We've got a big summer trip

[02:23:53] We've got a big summer trip coming up, and we've got to find

[02:23:56] coming up, and we've got to find a kennel for our two dogs, and

[02:23:57] a kennel for our two dogs, and here's a picture of our two

[02:23:59] here's a picture of our two dogs, there's Hank, looking

[02:24:01] dogs, there's Hank, looking good, and Louis Cinnamon, one of

[02:24:03] good, and Louis Cinnamon, one of the most interesting names for a

[02:24:04] the most interesting names for a dog we've ever heard.

[02:24:07] dog we've ever heard. What we're going to do and

[02:24:08] What we're going to do and remember when you have to go to

[02:24:09] remember when you have to go to a new kennel there's lots of

[02:24:11] a new kennel there's lots of paperwork, allergies, vaccines,

[02:24:13] paperwork, allergies, vaccines, all the history you have to pull

[02:24:15] all the history you have to pull together. It's so painful. And

[02:24:17] together. It's so painful. And so what you can do with this on

[02:24:20] so what you can do with this on Gemini, on Mac OS is actually

[02:24:24] Gemini, on Mac OS is actually take a look at a bunch of

[02:24:25] take a look at a bunch of documents like this. You'll be

[02:24:26] documents like this. You'll be able to select them all and

[02:24:28] able to select them all and long-press the Function Key and

[02:24:30] long-press the Function Key and just dictate the e-mail to the

[02:24:32] just dictate the e-mail to the kennel.

[02:24:32] kennel. So it works something like this.

[02:24:36] So it works something like this. Hi, there, I need to do a short

[02:24:39] Hi, there, I need to do a short boarding stay for my two dogs,

[02:24:41] boarding stay for my two dogs, Louis Cinnamon and Hank starting

[02:24:43] Louis Cinnamon and Hank starting this Thursday, oh, wait no

[02:24:44] this Thursday, oh, wait no actually, it's this Friday.

[02:24:46] actually, it's this Friday. They've never stayed with you

[02:24:47] They've never stayed with you before, but they're very social

[02:24:49] before, but they're very social dogs and also can you turn these

[02:24:51] dogs and also can you turn these files into a table with their

[02:24:55] files into a table with their details, allergies, recent

[02:24:56] details, allergies, recent vaccines and make this e-mail

[02:24:57] vaccines and make this e-mail sound friendly so we make a good

[02:25:00] sound friendly so we make a good first impression.

[02:25:02] first impression. All right. I'm going to release

[02:25:03] All right. I'm going to release the function key, you can see

[02:25:05] the function key, you can see Gemini is thinking at the bottom

[02:25:06] Gemini is thinking at the bottom here on this Macbook. What it's

[02:25:10] here on this Macbook. What it's done is because I've selected

[02:25:11] done is because I've selected those files in Finder, using its

[02:25:14] those files in Finder, using its multimodal understanding, it can

[02:25:15] multimodal understanding, it can go through the PDFs, it can go

[02:25:18] go through the PDFs, it can go through these images of their

[02:25:20] through these images of their invoices, and it's all

[02:25:21] invoices, and it's all controlled by my voice. So it

[02:25:23] controlled by my voice. So it can actually take all of that

[02:25:24] can actually take all of that complex information and look at

[02:25:26] complex information and look at that, there it is. It's got a

[02:25:29] that, there it is. It's got a table in line --

[02:25:31] table in line -- [Applause]

[02:25:34] [Applause] It's also so amazing, because it

[02:25:36] It's also so amazing, because it corrects -- remember, I said

[02:25:37] corrects -- remember, I said Thursday, no scratch that

[02:25:39] Thursday, no scratch that Friday, and it Pics that up and

[02:25:41] Friday, and it Pics that up and automatically cleans up my

[02:25:43] automatically cleans up my input.

[02:25:43] input. This is the power of what Gemini

[02:25:45] This is the power of what Gemini can do using your voice.

[02:25:49] can do using your voice. These new voice capabilities in

[02:25:53] These new voice capabilities in Gemini Spark will be coming to

[02:25:54] Gemini Spark will be coming to the Mac app this summer, as

[02:25:55] the Mac app this summer, as well.

[02:25:56] well. [Applause]

[02:26:00] [Applause] And so there it is. Today has

[02:26:01] And so there it is. Today has been a jam-packed Gemini day.

[02:26:04] been a jam-packed Gemini day. We've completely redesigned the

[02:26:07] We've completely redesigned the entire experience with Neural

[02:26:10] entire experience with Neural Expressive. We're shipping the

[02:26:11] Expressive. We're shipping the brand-new Gemini Omni model and

[02:26:12] brand-new Gemini Omni model and three the .5 Flash model with

[02:26:17] three the .5 Flash model with 3.5 Pro coming soon and you can

[02:26:20] 3.5 Pro coming soon and you can now put Gemini to work. And it

[02:26:22] now put Gemini to work. And it will keep working for you even

[02:26:23] will keep working for you even while you sleep, thanks to new

[02:26:24] while you sleep, thanks to new features like the Daily Brief

[02:26:25] features like the Daily Brief and Gemini Spark.

[02:26:28] and Gemini Spark. All of this gets us closer to

[02:26:30] All of this gets us closer to our vision of a universal

[02:26:32] our vision of a universal assistant that's personal,

[02:26:35] assistant that's personal, proactive, and powerful in your

[02:26:36] proactive, and powerful in your daily life. So whether you're a

[02:26:38] daily life. So whether you're a student, a busy parent, or a

[02:26:40] student, a busy parent, or a small business owner, we look

[02:26:42] small business owner, we look forward to what you're going to

[02:26:43] forward to what you're going to be able to do with Gemini.

[02:26:44] be able to do with Gemini. Thank you.

[02:26:45] Thank you. [Applause]

[02:26:47] [Applause] >> Hi.

[02:26:48] >> Hi. I'm Holly.

[02:26:49] I'm Holly. I moved from South Korea to the

[02:26:52] I moved from South Korea to the U.S. in 2005.

[02:26:55] U.S. in 2005. Food has always been something

[02:26:56] Food has always been something we love to do in our family.

[02:26:57] we love to do in our family. So I decided to open a

[02:26:59] So I decided to open a restaurant.

[02:27:02] restaurant. But running a restaurant, so

[02:27:05] But running a restaurant, so hard.

[02:27:09] hard. I discovered Gemini could help

[02:27:12] I discovered Gemini could help me with a lot of things.

[02:27:15] me with a lot of things. From menus to marketing,

[02:27:16] From menus to marketing, budgeting and inventory.

[02:27:16] budgeting and inventory. How much sesame oil did I order

[02:27:17] How much sesame oil did I order last month?

[02:27:17] last month? >> Last month you ordered five

[02:27:18] >> Last month you ordered five gallons.

[02:27:22] gallons. >> I even rebuilt our website

[02:27:22] >> I even rebuilt our website was Antigravity and Stitch to

[02:27:25] was Antigravity and Stitch to include a custom chatbot for ou

[02:27:26] include a custom chatbot for ou customers.

[02:27:28] customers. But one day we needed to find a

[02:27:32] But one day we needed to find a dishwasher very last minute and

[02:27:33] dishwasher very last minute and that was my "aha" moment.

[02:27:39] that was my "aha" moment. I wanted to help small

[02:27:39] I wanted to help small businesses and people find each

[02:27:40] businesses and people find each other faster.

[02:27:41] other faster. So I got a small team together

[02:27:42] So I got a small team together and we used Gemini models to

[02:27:43] and we used Gemini models to build an inclusive hiring

[02:27:47] build an inclusive hiring platform called WorkOnward.

[02:27:48] platform called WorkOnward. And we did what other job sites

[02:27:49] And we did what other job sites are not doing.

[02:27:51] are not doing. We translated the job postings

[02:27:52] We translated the job postings and then made it possible to

[02:27:54] and then made it possible to post jobs just by text.

[02:27:56] post jobs just by text. It helps break down barriers fo

[02:27:58] It helps break down barriers fo those who aren't tech savvy and

[02:28:03] those who aren't tech savvy and helps people find their place.

[02:28:06] helps people find their place. Our little tool is now a

[02:28:08] Our little tool is now a platform used by over 13,000

[02:28:10] platform used by over 13,000 people in New York City.

[02:28:13] people in New York City. It's about honoring the dignity

[02:28:15] It's about honoring the dignity of workers and small businesses

[02:28:17] of workers and small businesses and I think that's a power of

[02:28:18] and I think that's a power of AI.

[02:28:21] AI. ♪

[02:28:21] ♪ ♪

[02:28:27] ♪ [Applause]

[02:28:36] [Applause] >> SUZ CHAMBER: As you've heard

[02:28:38] >> SUZ CHAMBER: As you've heard today, our models and products

[02:28:44] today, our models and products are unlocking new breakthroughs,

[02:28:45] are unlocking new breakthroughs, but the real breakthrough isn't

[02:28:47] but the real breakthrough isn't the technology; it's what you do

[02:28:49] the technology; it's what you do with it.

[02:28:51] with it. Whether you're a designer, an

[02:28:53] Whether you're a designer, an entrepreneur, or an artist, our

[02:28:54] entrepreneur, or an artist, our products help shrink the gap

[02:28:55] products help shrink the gap between the moment you have an

[02:28:58] between the moment you have an idea and the moment you create

[02:28:59] idea and the moment you create it. At its best, technology is

[02:29:00] it. At its best, technology is a canvas for human creativity,

[02:29:04] a canvas for human creativity, and today, I want to dive into

[02:29:07] and today, I want to dive into the three products that help you

[02:29:09] the three products that help you bring your ideas to life. Let's

[02:29:11] bring your ideas to life. Let's start with one that takes the

[02:29:13] start with one that takes the power of Nano Banana and gives

[02:29:14] power of Nano Banana and gives you even more creative control.

[02:29:18] you even more creative control. Introducing Google Pics, a new

[02:29:20] Introducing Google Pics, a new product in Google Workspace.

[02:29:23] product in Google Workspace. Pics is our image creation and

[02:29:24] Pics is our image creation and editing tool that helps you

[02:29:26] editing tool that helps you create just about anything, from

[02:29:31] create just about anything, from party flyers to infographics

[02:29:32] party flyers to infographics with the creative control you

[02:29:35] with the creative control you want. Watch how easy this is.

[02:29:36] want. Watch how easy this is. You start with a base image as

[02:29:38] You start with a base image as your canvas, and what's really

[02:29:39] your canvas, and what's really cool is that Pics understands

[02:29:42] cool is that Pics understands what's in your creations and how

[02:29:45] what's in your creations and how the objects work together. You

[02:29:46] the objects work together. You can hover over an element and

[02:29:48] can hover over an element and click to remove it, or you can

[02:29:50] click to remove it, or you can resize an object to fit the

[02:29:54] resize an object to fit the frame. Once the layout is set,

[02:29:56] frame. Once the layout is set, you can add or edit text and

[02:29:59] you can add or edit text and translate all of it with just a

[02:30:04] translate all of it with just a few clicks.

[02:30:07] few clicks. [Applause]

[02:30:07] [Applause] Pretty cool.

[02:30:09] Pretty cool. Every output from our creative

[02:30:13] Every output from our creative tools, including Pics, is

[02:30:14] tools, including Pics, is watermarked with SynthID and

[02:30:15] watermarked with SynthID and Pics is rolling out this summer.

[02:30:19] Pics is rolling out this summer. [Applause]

[02:30:24] [Applause] But what if you want to go

[02:30:26] But what if you want to go beyond images? Maybe you want

[02:30:28] beyond images? Maybe you want to design an app or a website.

[02:30:31] to design an app or a website. Now, you can build UI at the

[02:30:33] Now, you can build UI at the speed of thought. Teams across

[02:30:36] speed of thought. Teams across Google use our design product

[02:30:38] Google use our design product called Stitch to turn rough

[02:30:41] called Stitch to turn rough ideas into beautiful UI designs.

[02:30:43] ideas into beautiful UI designs. But it's not just us. In the

[02:30:45] But it's not just us. In the last year, the world used Stitch

[02:30:48] last year, the world used Stitch to generate over 100 million UI

[02:30:52] to generate over 100 million UI screens. And to continue that

[02:30:53] screens. And to continue that momentum, today, we're

[02:30:53] momentum, today, we're introducing new ways to design.

[02:30:55] introducing new ways to design. Let's take a look at how this

[02:30:56] Let's take a look at how this works.

[02:31:00] works. My friends Tyler and Jenny own a

[02:31:01] My friends Tyler and Jenny own a pizza company. They have a ton

[02:31:06] pizza company. They have a ton of experience making pizza and

[02:31:06] of experience making pizza and no experience designing

[02:31:08] no experience designing websites. With a single prompt,

[02:31:09] websites. With a single prompt, Stitch will go to work,

[02:31:11] Stitch will go to work, generating that UI live.

[02:31:13] generating that UI live. Now, this is just the first

[02:31:16] Now, this is just the first pass, but if they want to refine

[02:31:19] pass, but if they want to refine it, they can collaborate with

[02:31:20] it, they can collaborate with Stitch in real time, either by

[02:31:21] Stitch in real time, either by writing prompts or using their

[02:31:22] writing prompts or using their voices. For example, they can

[02:31:27] voices. For example, they can jump in and say "make the header

[02:31:32] jump in and say "make the header text larger"and update the menu

[02:31:35] text larger"and update the menu to highlight more pizza

[02:31:36] to highlight more pizza options."

[02:31:41] options." And the layout updates in real

[02:31:41] And the layout updates in real time.

[02:31:48] time. And because Stitch links to many

[02:31:49] And because Stitch links to many tools, they can export their

[02:31:49] tools, they can export their design to code or launch their

[02:31:50] design to code or launch their website in just a few clicks.

[02:31:51] website in just a few clicks. These updates in Stitch are

[02:31:52] These updates in Stitch are rolling out today to users

[02:31:53] rolling out today to users globally.

[02:31:54] globally. [Applause]

[02:31:57] [Applause] With every new technology,

[02:32:01] With every new technology, what's most exciting is seeing

[02:32:03] what's most exciting is seeing what people make with it.

[02:32:04] what people make with it. That's why from day one, we

[02:32:05] That's why from day one, we haven't just built models and

[02:32:07] haven't just built models and tools for creatives; we've built

[02:32:08] tools for creatives; we've built with them. Let's take a look at

[02:32:10] with them. Let's take a look at some of our amazing partners and

[02:32:12] some of our amazing partners and what we've created together.

[02:32:19] what we've created together. >> You're in the era where the

[02:32:20] >> You're in the era where the human has to be the most

[02:32:21] human has to be the most creative.

[02:32:21] creative. ♪

[02:32:22] ♪ ♪

[02:32:33] ♪ >> Toto, I have a feeling we're

[02:32:35] >> Toto, I have a feeling we're not in Kansas anymore.

[02:32:39] not in Kansas anymore. >> Right?

[02:32:39] >> Right? ♪

[02:32:40] ♪ ♪

[02:32:54] ♪ >> Yeah. We'll do that.

[02:32:56] >> Yeah. We'll do that. ♪

[02:32:58] ♪ [Applause]

[02:33:03] [Applause] >> SUZ CHAMBER: That spirit of

[02:33:06] >> SUZ CHAMBER: That spirit of collaboration is why we launched

[02:33:08] collaboration is why we launched Google Flow at I/O last year.

[02:33:11] Google Flow at I/O last year. Today, millions of people are

[02:33:12] Today, millions of people are using it to create images,

[02:33:16] using it to create images, films, and music, in ways that

[02:33:18] films, and music, in ways that never could have been done

[02:33:18] never could have been done before and to build on that

[02:33:19] before and to build on that progress, we're rolling out

[02:33:21] progress, we're rolling out Gemini Omni, a new agent, custom

[02:33:23] Gemini Omni, a new agent, custom tools, and music remixing

[02:33:31] tools, and music remixing [Applause]

[02:33:34] [Applause] Let's start with Gemini Omni.

[02:33:38] Let's start with Gemini Omni. Take a look at this raw footage.

[02:33:39] Take a look at this raw footage. I love how this person is

[02:33:42] I love how this person is walking, his presence, his

[02:33:44] walking, his presence, his pacing, let's not change any of

[02:33:47] pacing, let's not change any of that.

[02:33:47] that. With a simple prompt and style

[02:33:52] With a simple prompt and style reference, Omni allows us to

[02:33:56] reference, Omni allows us to transform the environment, add

[02:33:58] transform the environment, add visual effects and any other

[02:33:59] visual effects and any other element all while preserving the

[02:33:59] element all while preserving the original performance.

[02:34:01] original performance. And now, you can even add new

[02:34:05] And now, you can even add new characters, while maintaining

[02:34:06] characters, while maintaining everything else in the scene.

[02:34:11] everything else in the scene. [Applause]

[02:34:12] [Applause] Next, let's take a look at our

[02:34:14] Next, let's take a look at our second big update, a new agent

[02:34:15] second big update, a new agent in Google Flow.

[02:34:18] in Google Flow. Until today, Flow could only

[02:34:23] Until today, Flow could only execute one prompt at a time.

[02:34:23] execute one prompt at a time. Now, your agent can take

[02:34:24] Now, your agent can take multiple actions all at once.

[02:34:29] multiple actions all at once. Starting with just a single

[02:34:33] Starting with just a single image, I can ask the agent to

[02:34:34] image, I can ask the agent to help me find the best camera

[02:34:35] help me find the best camera angles for this scene. It

[02:34:36] angles for this scene. It analyzes what's happening in the

[02:34:39] analyzes what's happening in the image, concepts the most

[02:34:43] image, concepts the most compelling angles and then boom,

[02:34:45] compelling angles and then boom, a single image becomes 16 unique

[02:34:50] a single image becomes 16 unique videos.

[02:34:51] videos. [Applause]

[02:34:52] [Applause] The agent can also handle

[02:34:55] The agent can also handle large-scale edits, like

[02:34:57] large-scale edits, like transforming all of these scenes

[02:34:59] transforming all of these scenes from early morning to late at

[02:35:03] from early morning to late at night.

[02:35:05] night. Its understanding of context is

[02:35:06] Its understanding of context is precise. The desert sky goes

[02:35:08] precise. The desert sky goes completely dark, and the

[02:35:11] completely dark, and the headlights turn on illuminating

[02:35:12] headlights turn on illuminating the dust. It's a true

[02:35:13] the dust. It's a true collaborator, helping you create

[02:35:15] collaborator, helping you create and edit at scale. Our next

[02:35:19] and edit at scale. Our next --

[02:35:19] -- [Applause]

[02:35:25] [Applause] Our next update is Flow tools.

[02:35:25] Our next update is Flow tools. Now, you can vibe-code any

[02:35:28] Now, you can vibe-code any creative tool you can think of,

[02:35:30] creative tool you can think of, right in Flow. Custom built by

[02:35:33] right in Flow. Custom built by you for your unique creative

[02:35:35] you for your unique creative process, like designing video

[02:35:38] process, like designing video effects, hand-drawn animations

[02:35:39] effects, hand-drawn animations or layering text.

[02:35:41] or layering text. You can start building, sharing,

[02:35:43] You can start building, sharing, and remixing tools today.

[02:35:47] and remixing tools today. [Applause]

[02:35:53] [Applause] Visual magic is only half the

[02:35:54] Visual magic is only half the story. Google Flow Music brings

[02:35:56] story. Google Flow Music brings the same creative control to

[02:35:58] the same creative control to help artists create original

[02:35:58] help artists create original songs.

[02:36:09] songs. For months, one of our teammates

[02:36:10] For months, one of our teammates had a piano riff in his head.

[02:36:11] had a piano riff in his head. Let's listen to that original

[02:36:11] Let's listen to that original recording.

[02:36:12] recording. ♪ [ piano ] ♪

[02:36:22] ♪ [ piano ] ♪ It's a really cool foundation,

[02:36:24] It's a really cool foundation, but he wanted to turn it into a

[02:36:25] but he wanted to turn it into a demo to guide his band. So he

[02:36:27] demo to guide his band. So he recorded his piano into Flow

[02:36:28] recorded his piano into Flow Music and prompted it for an R&B

[02:36:29] Music and prompted it for an R&B direction with a female vocal to

[02:36:30] direction with a female vocal to inspire his band's singer.

[02:36:31] inspire his band's singer. Let's take a listen.

[02:36:37] Let's take a listen. ♪

[02:36:38] ♪ ♪

[02:36:51] ♪ [Applause]

[02:36:59] [Applause] Now, this isn't his final track,

[02:37:00] Now, this isn't his final track, but it helped his band decide

[02:37:03] but it helped his band decide what to record next. These new

[02:37:07] what to record next. These new features in Google Flow and

[02:37:09] features in Google Flow and Google Flow Music are available

[02:37:11] Google Flow Music are available today.

[02:37:12] today. [Applause]

[02:37:21] [Applause] From musicians to small

[02:37:21] From musicians to small businesses, and vibe coders to

[02:37:22] businesses, and vibe coders to artists, the real breakthrough

[02:37:24] artists, the real breakthrough isn't the technology; it's what

[02:37:26] isn't the technology; it's what you do with it. And we can't

[02:37:27] you do with it. And we can't wait to see what you create

[02:37:31] wait to see what you create [Applause]

[02:37:35] [Applause] Next, I would like to hand

[02:37:37] Next, I would like to hand things over to Shahram to show

[02:37:38] things over to Shahram to show you what happens when we take

[02:37:39] you what happens when we take Google's latest innovations and

[02:37:41] Google's latest innovations and bring them into the real world.

[02:37:43] bring them into the real world. [Applause]

[02:37:43] [Applause] ♪

[02:37:44] ♪ ♪

[02:37:58] ♪ >> SHAHRAM IZADI: This is such

[02:38:00] >> SHAHRAM IZADI: This is such an exciting time for XR. AI

[02:38:05] an exciting time for XR. AI continues to unlock all-new

[02:38:05] continues to unlock all-new experiences on headsets, glasses

[02:38:07] experiences on headsets, glasses and everything in between.

[02:38:10] and everything in between. Android XR, the new platform

[02:38:12] Android XR, the new platform we've built with Samsung and

[02:38:14] we've built with Samsung and optimized for Snapdragon with

[02:38:17] optimized for Snapdragon with Qualcomm, combines this

[02:38:20] Qualcomm, combines this pioneering hardware with Gemini.

[02:38:25] pioneering hardware with Gemini. This gives you help in the

[02:38:27] This gives you help in the moment without taking you out of

[02:38:27] moment without taking you out of it. The next big milestone for

[02:38:29] it. The next big milestone for Android XR is Intelligent

[02:38:32] Android XR is Intelligent Eyewear. There will be two

[02:38:33] Eyewear. There will be two types of these AI glasses that

[02:38:35] types of these AI glasses that connect to your phone and give

[02:38:38] connect to your phone and give you hands-free help all day

[02:38:40] you hands-free help all day long.

[02:38:40] long. Last year, we showed you display

[02:38:47] Last year, we showed you display glasses on the I/O stage. With

[02:38:48] glasses on the I/O stage. With the small in-lens display,

[02:38:49] the small in-lens display, you'll get helpful information

[02:38:51] you'll get helpful information right in front of you, right

[02:38:55] right in front of you, right when you need it, like seeing

[02:38:55] when you need it, like seeing your Uber pickup details at a

[02:38:58] your Uber pickup details at a glance, or getting live

[02:38:59] glance, or getting live translations as you travel.

[02:39:01] translations as you travel. You'll even be able to use

[02:39:03] You'll even be able to use features like Create My Widget

[02:39:06] features like Create My Widget to make glance-able elements.

[02:39:09] to make glance-able elements. The first wave of developers are

[02:39:11] The first wave of developers are already creating display

[02:39:13] already creating display experiences, and you'll hear

[02:39:18] experiences, and you'll hear more about these glasses later

[02:39:19] more about these glasses later this year when we expand our

[02:39:20] this year when we expand our Trusted Tester Program. But

[02:39:22] Trusted Tester Program. But let's talk about what's

[02:39:26] let's talk about what's launching this year.

[02:39:27] launching this year. Today, I'm excited to announce

[02:39:29] Today, I'm excited to announce that our first audio glasses

[02:39:34] that our first audio glasses will arrive this fall.

[02:39:35] will arrive this fall. [Applause]

[02:39:38] [Applause] They are designed to give you

[02:39:40] They are designed to give you all-day help from Gemini that is

[02:39:45] all-day help from Gemini that is spoken into your ear privately

[02:39:45] spoken into your ear privately rather than shown on a display.

[02:39:48] rather than shown on a display. And these glasses let you stay

[02:39:50] And these glasses let you stay hands-free and heads-up for

[02:39:53] hands-free and heads-up for things like listening to music,

[02:39:55] things like listening to music, taking photos, making calls, or

[02:39:57] taking photos, making calls, or tapping into your phone apps

[02:40:00] tapping into your phone apps without reaching into your

[02:40:02] without reaching into your pocket.

[02:40:03] pocket. Personally, I love to cook, but

[02:40:04] Personally, I love to cook, but I'm not one to follow a recipe

[02:40:07] I'm not one to follow a recipe book, so it's great to have

[02:40:08] book, so it's great to have Gemini offer up some advice

[02:40:11] Gemini offer up some advice before I get too experimental.

[02:40:12] before I get too experimental. These audio glasses have brought

[02:40:16] These audio glasses have brought together an all-star cast of

[02:40:17] together an all-star cast of partners. You've got the

[02:40:20] partners. You've got the world's top eyewear designers at

[02:40:22] world's top eyewear designers at Gentle Monster and Warby Parker

[02:40:24] Gentle Monster and Warby Parker creating iconic designs. The

[02:40:28] creating iconic designs. The --

[02:40:29] -- [Applause]

[02:40:29] [Applause] Thank you.

[02:40:34] Thank you. The world's leading electronics

[02:40:36] The world's leading electronics company, Samsung, is building

[02:40:37] company, Samsung, is building innovative new devices and

[02:40:38] innovative new devices and experiences that set the bar for

[02:40:44] experiences that set the bar for the whole industry. And we've

[02:40:45] the whole industry. And we've been working to bring the best

[02:40:48] been working to bring the best of Google to these glasses, as

[02:40:51] of Google to these glasses, as well.

[02:40:54] well. And yes. They're going to pair

[02:40:54] And yes. They're going to pair with Android and iOS devices.

[02:40:55] with Android and iOS devices. [Applause]

[02:40:59] [Applause] They look incredible and today,

[02:41:02] They look incredible and today, You are actually going to

[02:41:03] You are actually going to finally see for yourself.

[02:41:06] finally see for yourself. Let's pass it to our friend and

[02:41:07] Let's pass it to our friend and partner from Samsung, Jay Kim,

[02:41:08] partner from Samsung, Jay Kim, to kick off the world's first

[02:41:10] to kick off the world's first look.

[02:41:16] look. >> At Samsung, our vision is to

[02:41:17] >> At Samsung, our vision is to enrich people's lives and help

[02:41:19] enrich people's lives and help shape how we live tomorrow.

[02:41:24] shape how we live tomorrow. In close partnership with

[02:41:25] In close partnership with Google,

[02:41:26] Google, we're introducing intelligent

[02:41:27] we're introducing intelligent eyewear that empowers you to

[02:41:28] eyewear that empowers you to connect to the world with

[02:41:28] connect to the world with confidence.

[02:41:29] confidence. Built with Samsung's precise

[02:41:30] Built with Samsung's precise engineering and craftsmanship,

[02:41:36] engineering and craftsmanship, we're merging form, function an

[02:41:36] we're merging form, function an helpful intelligence to create

[02:41:38] helpful intelligence to create something you'll want to wear.

[02:41:38] something you'll want to wear. In eyewear, every millimeter

[02:41:41] In eyewear, every millimeter counts.

[02:41:41] counts. Today, we're thrilled to

[02:41:43] Today, we're thrilled to share a first look at the

[02:41:45] share a first look at the upcoming styles co-created with

[02:41:50] upcoming styles co-created with our eyewear partners, Warby

[02:41:50] our eyewear partners, Warby Parker and Gentle Monster.

[02:41:51] Parker and Gentle Monster. Let's take a look.

[02:41:59] Let's take a look. >> People are always saying,

[02:41:59] >> People are always saying, “Hey, we have to be

[02:42:01] “Hey, we have to be disruptive.” But we have to

[02:42:04] disruptive.” But we have to think about what is the

[02:42:05] think about what is the disruption?

[02:42:05] disruption? I want to make intelligent

[02:42:06] I want to make intelligent eyewear that looks prettier tha

[02:42:06] eyewear that looks prettier tha normal eyewear.

[02:42:08] normal eyewear. That's our goal.

[02:42:09] That's our goal. It's the balance between

[02:42:12] It's the balance between technology and fashion.

[02:42:13] technology and fashion. It's not only product.

[02:42:14] It's not only product. It's all about perception and

[02:42:15] It's all about perception and emotion.

[02:42:19] emotion. You know, heart, that is the

[02:42:19] You know, heart, that is the DNA.

[02:42:19] DNA. I want to give confidence to

[02:42:20] I want to give confidence to people.

[02:42:23] people. When they wear this, they feel

[02:42:24] When they wear this, they feel like connected with this kind o

[02:42:28] like connected with this kind o braveness, these kind of

[02:42:28] braveness, these kind of rebellious things because this

[02:42:29] rebellious things because this is who we are.

[02:42:30] is who we are. That's the meaning of Gentle

[02:42:31] That's the meaning of Gentle Monster with Google and Samsung

[02:42:34] Monster with Google and Samsung >> Glasses are a deeply persona

[02:42:37] >> Glasses are a deeply persona product.

[02:42:38] product. They shape how you see the worl

[02:42:39] They shape how you see the worl and how the world sees you.

[02:42:43] and how the world sees you. >> What you see here is an

[02:42:44] >> What you see here is an evolution of Warby Parker's

[02:42:45] evolution of Warby Parker's first intelligent eyewear

[02:42:47] first intelligent eyewear designs.

[02:42:50] designs. >> Our vision for intelligent

[02:42:51] >> Our vision for intelligent eyewear started by wanting to

[02:42:51] eyewear started by wanting to design a beautiful pair of

[02:42:52] design a beautiful pair of glasses.

[02:42:54] glasses. We often take inspiration from

[02:42:55] We often take inspiration from different art objects, artists,

[02:42:57] different art objects, artists, eras in time.

[02:43:01] eras in time. >> It was important for us not

[02:43:02] >> It was important for us not to hide the technology but to

[02:43:03] to hide the technology but to celebrate it.

[02:43:05] celebrate it. >> These will not only help

[02:43:06] >> These will not only help people see but will help them

[02:43:07] people see but will help them understand the world more

[02:43:09] understand the world more deeply.

[02:43:10] deeply. >> We want people to experience

[02:43:11] >> We want people to experience the world fully and that's the

[02:43:12] the world fully and that's the magic of this technology.

[02:43:15] magic of this technology. Hey, Gemini, who wore it better

[02:43:17] Hey, Gemini, who wore it better [Applause]

[02:43:28] [Applause] >> SHAHRAM IZADI: Aren't they

[02:43:36] >> SHAHRAM IZADI: Aren't they beautiful? These are the firs

[02:43:39] beautiful? These are the firs Two designs of a bigger

[02:43:40] Two designs of a bigger collection coming this fall.

[02:43:42] collection coming this fall. You know what time it is now.

[02:43:43] You know what time it is now. Who's up for a live demo?

[02:43:45] Who's up for a live demo? [Applause]

[02:43:45] [Applause] Nishtha is going to join me on

[02:43:47] Nishtha is going to join me on stage, so let's have Gemini play

[02:43:49] stage, so let's have Gemini play some entrance music for her.

[02:43:52] some entrance music for her. And it's seeing the world around

[02:43:54] And it's seeing the world around you, you can see, so the more

[02:43:56] you, you can see, so the more hyped you are in the audience,

[02:43:57] hyped you are in the audience, the more pumping the music will

[02:44:01] the more pumping the music will be.

[02:44:06] be. Gemini... I think it's done it

[02:44:09] Gemini... I think it's done it already. Play that music! Come

[02:44:10] already. Play that music! Come on let's go.

[02:44:11] on let's go. ♪

[02:44:12] ♪ ♪

[02:44:15] ♪ Nishtha, welcome.

[02:44:20] Nishtha, welcome. Welcome, Nishtha.

[02:44:21] Welcome, Nishtha. >> NISHTHA BHATIA: Thank you.

[02:44:22] >> NISHTHA BHATIA: Thank you. >> SHAHRAM IZADI: I'm wearing

[02:44:23] >> SHAHRAM IZADI: I'm wearing one of the Warby Parker styles

[02:44:24] one of the Warby Parker styles and Nishtha has a pair of the

[02:44:25] and Nishtha has a pair of the Gentle Monster glasses on.

[02:44:27] Gentle Monster glasses on. Looking great, Nishtha.

[02:44:29] Looking great, Nishtha. >> NISHTHA BHATIA: Thanks, Ram,

[02:44:30] >> NISHTHA BHATIA: Thanks, Ram, you, too.

[02:44:32] you, too. >> SHAHRAM IZADI: A quick

[02:44:32] >> SHAHRAM IZADI: A quick preamble for the demo today.

[02:44:35] preamble for the demo today. We're going to be piping in the

[02:44:36] We're going to be piping in the audio to the stage speakers from

[02:44:37] audio to the stage speakers from our glasses, but in regular use,

[02:44:41] our glasses, but in regular use, of course, Gemini is going to

[02:44:42] of course, Gemini is going to just privately be talking to us.

[02:44:46] just privately be talking to us. Now, last time we were on stag

[02:44:50] Now, last time we were on stag We saw the rich Google Maps

[02:44:51] We saw the rich Google Maps experience on display glasses.

[02:44:53] experience on display glasses. Well Maps on these glasses are

[02:44:54] Well Maps on these glasses are just as helpful, especially when

[02:44:55] just as helpful, especially when working in combination with

[02:44:56] working in combination with Personal Intelligence.

[02:44:57] Personal Intelligence. >> NISHTHA BHATIA: No pulling

[02:45:01] >> NISHTHA BHATIA: No pulling out my phone. All I have to do

[02:45:05] out my phone. All I have to do Now is ask.

[02:45:09] Now is ask. Can you navigate me to that

[02:45:09] Can you navigate me to that place I met my friend Gianna at

[02:45:10] place I met my friend Gianna at last week?

[02:45:12] last week? >> GEMINI: Hey, Nishtha, I've

[02:45:13] >> GEMINI: Hey, Nishtha, I've set your route to the Redwood

[02:45:13] set your route to the Redwood Grove Natural Preserve from Last

[02:45:19] Grove Natural Preserve from Last Week's Hike, want to add a stop

[02:45:20] Week's Hike, want to add a stop on the way to grab your

[02:45:21] on the way to grab your afternoon cold brew?

[02:45:21] afternoon cold brew? >> NISHTHA BHATIA: Yes, Gemini I

[02:45:22] >> NISHTHA BHATIA: Yes, Gemini I Would Love That.

[02:45:23] Would Love That. >> GEMINI: Okay. I'm Starting

[02:45:24] >> GEMINI: Okay. I'm Starting Walking Navigation with a Stop

[02:45:24] Walking Navigation with a Stop at Koopa Cafe. Turn Around and

[02:45:25] at Koopa Cafe. Turn Around and Head Towards Bill Graham

[02:45:27] Head Towards Bill Graham Parkway.

[02:45:29] Parkway. >> SHAHRAM IZADI: With glasses,

[02:45:29] >> SHAHRAM IZADI: With glasses, you can allow Maps to further

[02:45:33] you can allow Maps to further understand your context, and

[02:45:34] understand your context, and what's in front of you so you

[02:45:35] what's in front of you so you get detailed directions like the

[02:45:37] get detailed directions like the coffee shop is coming up on your

[02:45:38] coffee shop is coming up on your right.

[02:45:39] right. Speaking of coffee, Nishtha, do

[02:45:40] Speaking of coffee, Nishtha, do you want to share how Gemini

[02:45:42] you want to share how Gemini intelligence can allow you to

[02:45:43] intelligence can allow you to order that cold brew up ahead?

[02:45:47] order that cold brew up ahead? >> NISHTHA BHATIA: That's a

[02:45:49] >> NISHTHA BHATIA: That's a really good idea. Gemini, can

[02:45:50] really good idea. Gemini, can you actually put my usual order

[02:45:52] you actually put my usual order in at that coffee shop we just

[02:45:53] in at that coffee shop we just talked about?

[02:45:57] talked about? >> GEMINI: Sure, I'll order you

[02:45:57] >> GEMINI: Sure, I'll order you a nitro cold brew for pickup

[02:45:58] a nitro cold brew for pickup from Koopa Cafe on Doordash.

[02:46:03] from Koopa Cafe on Doordash. >> SHAHRAM IZADI: On the screen

[02:46:04] >> SHAHRAM IZADI: On the screen behind me you can see Nishtha's

[02:46:06] behind me you can see Nishtha's phone in her pocket and Gemini

[02:46:07] phone in her pocket and Gemini is able to launch apps like

[02:46:12] is able to launch apps like Doordash and then click through

[02:46:12] Doordash and then click through all the different option screens

[02:46:18] all the different option screens automatically to order her

[02:46:18] automatically to order her coffee.

[02:46:19] coffee. And in a moment, it will be

[02:46:20] And in a moment, it will be ready for Nishtha's

[02:46:23] ready for Nishtha's confirmation.

[02:46:25] confirmation. There we go.

[02:46:27] There we go. [Applause]

[02:46:31] [Applause] There we go.

[02:46:35] There we go. >> GEMINI: I have prepared your

[02:46:37] >> GEMINI: I have prepared your order for the nitro cold brew

[02:46:39] order for the nitro cold brew from Koopa Cafe. Would you like

[02:46:40] from Koopa Cafe. Would you like to confirm?

[02:46:41] to confirm? >> NISHTHA BHATIA: Yes, please.

[02:46:43] >> NISHTHA BHATIA: Yes, please. And add a tip for 20%, too.

[02:46:47] And add a tip for 20%, too. >> SHAHRAM IZADI: I think after

[02:46:48] >> SHAHRAM IZADI: I think after all these rehearsals, that's her

[02:46:50] all these rehearsals, that's her 15th cold brew of the day and

[02:46:51] 15th cold brew of the day and thanks to the coffee shop for

[02:46:53] thanks to the coffee shop for putting up with us.

[02:46:55] putting up with us. >> NISHTHA BHATIA: Definitely,

[02:46:56] >> NISHTHA BHATIA: Definitely, is and speaking of preparation,

[02:46:57] is and speaking of preparation, I actually muted my texts before

[02:46:58] I actually muted my texts before I came on stage, but let me see

[02:47:00] I came on stage, but let me see if I can catch up now.

[02:47:02] if I can catch up now. Hey, Gemini, any important

[02:47:04] Hey, Gemini, any important messages I missed?

[02:47:08] messages I missed? >> GEMINI: Yes. Your family

[02:47:08] >> GEMINI: Yes. Your family group chat decided to meet for

[02:47:09] group chat decided to meet for dinner tonight at 7:00 and

[02:47:11] dinner tonight at 7:00 and there's a reminder to say your

[02:47:12] there's a reminder to say your daily affirmations. You are

[02:47:13] daily affirmations. You are strong --

[02:47:16] strong -- >> NISHTHA BHATIA: Okay. We'll

[02:47:17] >> NISHTHA BHATIA: Okay. We'll do the affirmations a bit later,

[02:47:19] do the affirmations a bit later, but why don't you add that

[02:47:20] but why don't you add that dinner to my calendar?

[02:47:24] dinner to my calendar? >> GEMINI: Sure I'll add an

[02:47:25] >> GEMINI: Sure I'll add an event for family dinner at

[02:47:26] event for family dinner at 7:00 p.m. to your calendar.

[02:47:26] 7:00 p.m. to your calendar. Right after your team

[02:47:27] Right after your team celebration.

[02:47:29] celebration. >> SHAHRAM IZADI: Awesome.

[02:47:30] >> SHAHRAM IZADI: Awesome. [Applause]

[02:47:32] [Applause] Thank you.

[02:47:35] Thank you. Gemini not only summarized her

[02:47:39] Gemini not only summarized her muted texts but it tapped into

[02:47:41] muted texts but it tapped into her Calendar app to add that

[02:47:41] her Calendar app to add that event.

[02:47:45] event. So we saw how glasses work with

[02:47:47] So we saw how glasses work with Your phone, but what about when

[02:47:48] Your phone, but what about when they attach to your watch to

[02:47:50] they attach to your watch to give you a glancable display?

[02:47:54] give you a glancable display? >> NISHTHA BHATIA: It's not a

[02:47:55] >> NISHTHA BHATIA: It's not a Google I/O without an audience

[02:47:55] Google I/O without an audience selfie, but we're going to do it

[02:47:56] selfie, but we're going to do it with a twist this time. Can

[02:47:56] with a twist this time. Can everyone strike their favorite

[02:47:58] everyone strike their favorite pose?

[02:48:00] pose? Gemini, take a photo of this

[02:48:01] Gemini, take a photo of this amazing audience but turn it

[02:48:03] amazing audience but turn it into a cartoon and add a big

[02:48:05] into a cartoon and add a big blimp in the sky that says

[02:48:06] blimp in the sky that says Google I/O 2026 on it in fun

[02:48:07] Google I/O 2026 on it in fun colors.

[02:48:10] colors. >> SHAHRAM IZADI: Everyone

[02:48:11] >> SHAHRAM IZADI: Everyone strike a pose.

[02:48:13] strike a pose. Okay. If this goes well,

[02:48:15] Okay. If this goes well, everyone go bananas in the

[02:48:19] everyone go bananas in the audience. Nano Banana on

[02:48:20] audience. Nano Banana on glasses is just awesome and just

[02:48:22] glasses is just awesome and just in a few seconds, you'll even

[02:48:24] in a few seconds, you'll even see that seamless preview on her

[02:48:28] see that seamless preview on her watch any second now, drum roll.

[02:48:31] watch any second now, drum roll. Here we go!

[02:48:33] Here we go! [Applause]

[02:48:43] [Applause] Thank you, Nishtha. Round of

[02:48:45] Thank you, Nishtha. Round of applause for Nishtha.

[02:48:46] applause for Nishtha. [Applause]

[02:48:49] [Applause] The demo worked, yes!

[02:48:53] The demo worked, yes! The future of Intelligent

[02:48:54] The future of Intelligent Eyewear has never been more

[02:48:56] Eyewear has never been more exciting. Stunning designs from

[02:48:59] exciting. Stunning designs from iconic brands, engineering and

[02:49:01] iconic brands, engineering and craftsmanship from Samsung,

[02:49:03] craftsmanship from Samsung, personal and proactive help from

[02:49:06] personal and proactive help from Gemini, customized apps and

[02:49:08] Gemini, customized apps and features from Google, and the

[02:49:13] features from Google, and the developer ecosystem, all

[02:49:16] developer ecosystem, all Arriving with our first glasses

[02:49:18] Arriving with our first glasses this fall. So stay tuned.

[02:49:20] this fall. So stay tuned. Now, I'm going to pass it to

[02:49:21] Now, I'm going to pass it to Demis.

[02:49:22] Demis. Thank you.

[02:49:23] Thank you. [Applause]

[02:49:27] [Applause] I'm going to pass it to Demis to

[02:49:28] I'm going to pass it to Demis to talk about the future of AI.

[02:49:29] talk about the future of AI. ♪

[02:49:30] ♪ ♪

[02:49:34] ♪ >> DEMIS HASSABIS: It's amazing

[02:49:36] >> DEMIS HASSABIS: It's amazing to see how far we've come with

[02:49:37] to see how far we've come with glasses, and I can't wait for

[02:49:39] glasses, and I can't wait for everyone to experience them.

[02:49:41] everyone to experience them. Today, we showed you our

[02:49:43] Today, we showed you our next-generation models, Gemini

[02:49:45] next-generation models, Gemini 3.5 and Omni, new coding

[02:49:48] 3.5 and Omni, new coding capabilities in Antigravity 2.0,

[02:49:49] capabilities in Antigravity 2.0, agents in Search and Gemini

[02:49:51] agents in Search and Gemini Spark, and so much more. It's

[02:49:55] Spark, and so much more. It's great to see Gemini transforming

[02:49:57] great to see Gemini transforming so many Google products used by

[02:50:00] so many Google products used by billions of people every day.

[02:50:02] billions of people every day. All of these advances show the

[02:50:03] All of these advances show the staggering pace of AI progress.

[02:50:04] staggering pace of AI progress. It's incredible, even for those

[02:50:06] It's incredible, even for those of us who have spent our entire

[02:50:08] of us who have spent our entire lives working on this. AGI is

[02:50:11] lives working on this. AGI is now on the horizon and it will

[02:50:12] now on the horizon and it will be the most profound and

[02:50:13] be the most profound and impactful technology ever

[02:50:15] impactful technology ever invented.

[02:50:19] invented. If built right, it could propel

[02:50:20] If built right, it could propel human progress and flourishing

[02:50:23] human progress and flourishing beyond our imagination. We're

[02:50:24] beyond our imagination. We're in a moment of immense promise,

[02:50:25] in a moment of immense promise, but also enormous

[02:50:27] but also enormous responsibility.

[02:50:28] responsibility. It's important that we are

[02:50:30] It's important that we are clear-eyed about the potential

[02:50:31] clear-eyed about the potential challenges and use all the tools

[02:50:35] challenges and use all the tools at our disposal to ensure the

[02:50:37] at our disposal to ensure the safety of our agentic systems

[02:50:39] safety of our agentic systems and ultimately, AGI itself.

[02:50:40] and ultimately, AGI itself. One area of risk that has gained

[02:50:42] One area of risk that has gained a lot of attention recently is

[02:50:43] a lot of attention recently is cybersecurity. Google has

[02:50:45] cybersecurity. Google has invested in this area for

[02:50:46] invested in this area for decades and we are bringing our

[02:50:47] decades and we are bringing our frontier capabilities and deep

[02:50:49] frontier capabilities and deep expertise to help secure the

[02:50:53] expertise to help secure the world's codebases.

[02:50:54] world's codebases. We have tools like our Code

[02:50:57] We have tools like our Code Security Agent, CodeMender,

[02:50:58] Security Agent, CodeMender, which automatically finds and

[02:50:59] which automatically finds and fixes critical software

[02:51:01] fixes critical software vulnerabilities. Today, we're

[02:51:05] vulnerabilities. Today, we're inviting a select group of

[02:51:05] inviting a select group of experts to test a new CodeMender

[02:51:06] experts to test a new CodeMender API, and we'll be launching it

[02:51:06] API, and we'll be launching it more broadly soon.

[02:51:11] more broadly soon. [Applause]

[02:51:18] [Applause] Stepping back, the whole reason

[02:51:18] Stepping back, the whole reason I've worked on AI my entire

[02:51:19] I've worked on AI my entire career was, because I saw it as

[02:51:20] career was, because I saw it as the ultimate tool to advance

[02:51:21] the ultimate tool to advance science and our understanding of

[02:51:23] science and our understanding of the world. It's awesome to see

[02:51:25] the world. It's awesome to see those dreams become a reality

[02:51:27] those dreams become a reality with AI beginning to help

[02:51:28] with AI beginning to help scientists in almost every

[02:51:30] scientists in almost every field.

[02:51:34] field. Building on this momentum, I'm

[02:51:35] Building on this momentum, I'm excited to announce Gemini for

[02:51:41] excited to announce Gemini for Science.

[02:51:42] Science. [Applause]

[02:51:45] [Applause] It brings together a number of

[02:51:46] It brings together a number of powerful AI tools to help

[02:51:48] powerful AI tools to help accelerate research.

[02:51:49] accelerate research. Gemini can already assist in

[02:51:51] Gemini can already assist in solving complex problems, but

[02:51:53] solving complex problems, but our new Labs prototypes

[02:51:54] our new Labs prototypes streamline daily scientific

[02:51:56] streamline daily scientific tasks, whether it's staying on

[02:52:00] tasks, whether it's staying on top of newly published papers,

[02:52:01] top of newly published papers, transforming research goals into

[02:52:02] transforming research goals into usable code or generating new

[02:52:05] usable code or generating new hypotheses.

[02:52:06] hypotheses. Another powerful tool for

[02:52:09] Another powerful tool for science is simulation. AI

[02:52:12] science is simulation. AI simulations are going to be

[02:52:14] simulations are going to be critical to understanding and

[02:52:16] critical to understanding and predicting dynamic systems that

[02:52:17] predicting dynamic systems that are simply too complex to model

[02:52:19] are simply too complex to model directly today. An amazing

[02:52:19] directly today. An amazing example of this is AlphaEarth

[02:52:22] example of this is AlphaEarth Foundations. It's the closest

[02:52:24] Foundations. It's the closest thing we have to a digital twin

[02:52:24] thing we have to a digital twin of the planet that could help

[02:52:25] of the planet that could help address problems like

[02:52:29] address problems like deforestation and food security.

[02:52:31] deforestation and food security. Simulations are already proving

[02:52:32] Simulations are already proving extremely useful. Our

[02:52:35] extremely useful. Our state-of-the-art WeatherNext

[02:52:35] state-of-the-art WeatherNext models can predict hurricane

[02:52:37] models can predict hurricane paths faster and more accurately

[02:52:39] paths faster and more accurately than traditional systems. Let's

[02:52:40] than traditional systems. Let's take a look at how weather

[02:52:48] take a look at how weather helped during last year's

[02:52:49] helped during last year's hurricane season.

[02:52:49] hurricane season. >> Tropical storms and

[02:52:50] >> Tropical storms and hurricanes can change very

[02:52:51] hurricanes can change very quickly, which makes them more

[02:52:51] quickly, which makes them more challenging to predict than

[02:52:52] challenging to predict than other types of weather systems.

[02:52:55] other types of weather systems. >> At Google, we developed

[02:52:58] >> At Google, we developed WeatherNext a global weather

[02:52:59] WeatherNext a global weather forecasting AI model that is

[02:52:59] forecasting AI model that is also able to predict where

[02:53:00] also able to predict where hurricanes are going to go and

[02:53:01] hurricanes are going to go and how strong they're going to

[02:53:02] how strong they're going to become. In 2025, WeatherNext

[02:53:04] become. In 2025, WeatherNext predicted a category 5 hurricane

[02:53:06] predicted a category 5 hurricane striking Jamaica three days

[02:53:07] striking Jamaica three days early with greater accuracy than

[02:53:08] early with greater accuracy than previous models.

[02:53:11] previous models. >> It's going to cause

[02:53:14] >> It's going to cause catastrophic, life-threatening

[02:53:15] catastrophic, life-threatening damage.

[02:53:21] damage. >> Because of that early

[02:53:23] >> Because of that early warning, we were able to give

[02:53:25] warning, we were able to give that advance notice to the

[02:53:29] that advance notice to the public to say, move from certain

[02:53:31] public to say, move from certain areas, so it saved their lives.

[02:53:32] areas, so it saved their lives. >> WeatherNext was a really

[02:53:33] >> WeatherNext was a really valuable tool in helping us make

[02:53:34] valuable tool in helping us make these more accurate and

[02:53:36] these more accurate and aggressive forecasts for

[02:53:38] aggressive forecasts for Melissa. And I think going

[02:53:39] Melissa. And I think going forward, WeatherNext and other

[02:53:43] forward, WeatherNext and other AI models will become a part of

[02:53:44] AI models will become a part of our routine forecast toolkit

[02:53:44] our routine forecast toolkit here at the Hurricane Center.

[02:53:46] here at the Hurricane Center. [Applause]

[02:53:54] [Applause] >> DEMIS HASSABIS: In the

[02:53:55] >> DEMIS HASSABIS: In the future, we will be able to

[02:53:55] future, we will be able to simulate even more complex

[02:53:59] simulate even more complex emergent systems, perhaps even

[02:53:59] emergent systems, perhaps even virtual cells. Our biological

[02:54:03] virtual cells. Our biological models, like AlphaFold and

[02:54:05] models, like AlphaFold and AlphaGenome, have already become

[02:54:05] AlphaGenome, have already become standard research tools used by

[02:54:06] standard research tools used by millions of scientists around

[02:54:08] millions of scientists around the world to make important

[02:54:11] the world to make important advances in their fields. I

[02:54:14] advances in their fields. I like to call this "science at

[02:54:19] like to call this "science at digital speeds," both in terms

[02:54:20] digital speeds," both in terms of the speed of the solution and

[02:54:21] of the speed of the solution and its dissemination to the

[02:54:21] its dissemination to the researchers who can make use of

[02:54:22] researchers who can make use of it.

[02:54:22] it. I've always believed the number

[02:54:26] I've always believed the number I've always believed the number

[02:54:27] I've always believed the number one application of AI should be

[02:54:28] one application of AI should be to improve human health. At

[02:54:29] to improve human health. At Isomorphic Labs, we're modeling

[02:54:30] Isomorphic Labs, we're modeling molecular interactions to

[02:54:32] molecular interactions to massively accelerate the

[02:54:33] massively accelerate the development of new medicines,

[02:54:35] development of new medicines, supported by leading industry

[02:54:37] supported by leading industry partners. We're now in the

[02:54:38] partners. We're now in the pre-clinical stage with multiple

[02:54:43] pre-clinical stage with multiple projects, including potential

[02:54:43] projects, including potential treatments for immune disorders

[02:54:49] treatments for immune disorders and cancer.

[02:54:50] and cancer. [Applause]

[02:54:53] [Applause] Our mission is to reimagine the

[02:54:55] Our mission is to reimagine the drug discovery process with the

[02:54:57] drug discovery process with the goal of one day solving all

[02:55:00] goal of one day solving all disease. Something that would

[02:55:01] disease. Something that would have seemed impossible just a

[02:55:03] have seemed impossible just a few years ago but I truly

[02:55:05] few years ago but I truly believe is now within reach.

[02:55:10] believe is now within reach. Google's cutting-edge research

[02:55:11] Google's cutting-edge research and products will help unlock

[02:55:12] and products will help unlock AGI's incredible potential for

[02:55:13] AGI's incredible potential for the benefit of the entire world.

[02:55:17] the benefit of the entire world. When we look back at this time,

[02:55:18] When we look back at this time, I think we will realize that we

[02:55:20] I think we will realize that we were standing in the foothills

[02:55:22] were standing in the foothills of the singularity.

[02:55:24] of the singularity. It will be a profound moment for

[02:55:33] It will be a profound moment for humanity. This technology will

[02:55:34] humanity. This technology will be a force multiplier for human

[02:55:34] be a force multiplier for human ingenuity and usher in a new

[02:55:36] ingenuity and usher in a new golden age of scientific

[02:55:37] golden age of scientific discovery and progress,

[02:55:38] discovery and progress, improving the lives of everyone,

[02:55:39] improving the lives of everyone, everywhere. We look forward to

[02:55:40] everywhere. We look forward to building the future with all of

[02:55:41] building the future with all of you. Thank you and enjoy the

[02:55:43] you. Thank you and enjoy the rest of Google I/O.

[02:55:44] rest of Google I/O. [Applause]

[02:55:47] [Applause] ♪

[02:55:47] ♪ ♪

Cite this page

If you're using ChatGPT, Claude, Gemini, or another AI assistant, paste this URL into the chat:

https://youtube-transcript.ai/docs/google-i-o-26-keynote-d0xuzxzywp

The full transcript and summary on this page will be retrieved as context, so the assistant can answer questions about the video accurately.