# Google I/O '26 Keynote

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] ♪ ♪