# Fable 5 Is Back, and I Only Have Days To Prove Its ROI

https://www.youtube.com/watch?v=5AXq2G14gUw

[00:00] Fable 5 is back.
[00:03] I'm really happy about it.
[00:05] The only thing that I see on YouTube and other channels, whenever people talk about these models, I rarely see real use cases where they are getting applied in real life work.
[00:14] And that's why in this video I will show you the first thing that I did in my business the moment I got access to Fable again.
[00:20] And I can tell you already, Fable didn't disappoint.
[00:22] It is really insane.
[00:25] And I really hate the fact that we will only get access through API to Fable in the future because that's very expensive.
[00:32] And that's why I take this time where I have access to it to find the best return of investment moments where it's worth using Fable.
[00:39] So in the future I know when it's worth investing money using Fable through API and when to use Opus or Sonnet in other use cases.
[00:50] And that's exactly what I wanted to figure out.
[00:52] So let's dive into this.
[00:54] Before we get started, let's talk about what my business actually is.
[00:56] It is my iCole.
[01:00] This YouTube channel, the membership platform that we have, all the courses that we have, and so on.
[01:06] And now you can say, "Well, this is a creator's use case and so on."
[01:08] But the thing is, I'm a busy professional.
[01:11] I worked in corporate for over 8 years.
[01:13] I climbed the ladder there, and the core of our business is teaching productivity end to end in a tool agnostic sense.
[01:20] This is what iCole is all about.
[01:23] Together with my co-founder Paco Cantero, who built and grew four businesses in parallel, still working there with 70 people employed, we are applying AI in the real world.
[01:36] And in our podcast we talk a lot about his use cases, too, how he's leveraging AI.
[01:41] But what I'm sharing here applies to all of us in some way or form.
[01:43] You just need to think about the complexity of any projects that you have.
[01:47] Then you will understand the thought process that I had here and can apply it to your own use cases.
[01:53] So let's go.
[01:56] The first thing I couldn't wait to get my hands back on Fable was to make a thorough audit of
[02:01] our i quad journey courses that teach people productivity end to end starting with digital note-taking, PKM, task management, project management, and eventually automation like a pro.
[02:14] And all these courses, each individual lesson has one concept or a workflow that we describe in these lessons.
[02:21] And if we go into one of these lessons, you see each lesson is very deep when it comes to this.
[02:25] So, we have a magic slide that explains the lesson.
[02:31] We have these chapters that that can go up to 30,000 words in one lesson.
[02:36] We have the connections to other resources that are related to this concept.
[02:40] And we have that explainer videos that go deep into this lesson, too, to give another perspective on the same thing.
[02:48] And the thing is, we have full access to everything, obviously, because we own the data.
[02:52] We have the databases.
[02:54] It's not a pre-made membership platform.
[02:56] That's why we built it from the ground up to really build the best learning platform in the world for this specific
[03:02] use case, building productivity systems end to end.
[03:04] And this comes along with the growth assignments that people have, where per lesson, you have questions to help you reflect yourself.
[03:12] Each answer gives you a specific way to improve it.
[03:15] Each question goes deep into the why behind the question.
[03:21] And by going through all these questions, you will figure out the blind spots and the gaps in your productivity system.
[03:28] So, you then can perfectly go back into the lessons and dive much deeper into the vertical content that we have here, because going through all lessons end to end and trying to read it all at once, that makes no sense.
[03:40] We need to apply what we learn directly in our real life.
[03:42] And that's what this whole learning platform is there, that you really understand what's the most important thing for me now, and then I can really dig deeper into this.
[03:51] And the challenge here is that with this amount of content, we also want to give people the feeling of progress starting with note-taking PKM, going to task
[04:03] management, and so on.
[04:05] So, it should be only be backwards referred to any previous concepts or workflows so people don't get confused when they move forward through the course.
[04:13] So, it makes no sense to assume here in the first lesson of the whole journey that you know already about the concept in a project management like a pro, for example.
[04:24] That makes no sense, so we need to make things backwards.
[04:26] So, when I'm in a project management like a pro, I can refer back to the concept that people learned earlier.
[04:32] And that's the only way how we can build a golden thread.
[04:34] And that's what we did already manually, also using AI in the past, but I knew, and that was actually the thing.
[04:41] Just the day when they switched off Fable, I wanted to roll this, and I did it, and I want to show you now exactly behind the scenes how I made the prompt to do this, and then the output.
[04:53] So, if you follow me, you know that I'm using VS Code, and I'm using Claude Code in here.
[04:59] No matter if you're a non-coder, there is no single line of code appearing here, even though it looks
[05:04] complex.
[05:06] And this is the way how you should use Claude.
[05:08] Everything else is a waste of tokens and time in my opinion if you don't use terminal and you stick to the desktop app.
[05:13] This is the thing if you really want to get the most out of your Claude subscription.
[05:15] So, here is what I started.
[05:19] It was just, you know, not even a refined prompt, but when it comes to refining prompts, stay with me because I will give you another tip here in a moment.
[05:28] But in this case, I was just writing down what I want from exactly this.
[05:30] So, look through the courses, understand the lessons, how are they interconnected, are there any gaps, anything that we need to fill in, and so on.
[05:38] But the important thing here is I don't just want to get a list in here in this terminal telling me now what to do.
[05:42] I always let AI create a HTML page, a web page, to present the data to me, so it's easier for me to understand.
[05:55] And here I really wanted to visualize even the cross-connections.
[05:57] It has access to the transcripts and the videos that we can directly edit through AI.
[06:01] Not
[06:06] showing you in this video how this works, but that's how we can do it.
[06:09] So, I let it go through all the video lessons, too, in order to understand are there any conceptions, is there anything wrongly connected?
[06:18] And that's exactly what I did to here, and the result looks then like this.
[06:20] This is the output.
[06:23] And this is what you get with Fabel.
[06:25] And I knew when I did this with Opus, and we'll get to this in a moment, big difference between Fabel and Opus in this case.
[06:31] I knew in Fabel he gets the point holistically.
[06:33] Fabel is able to really look through everything and start connecting the dots much better than any model I used before, and that was the huge potential I saw the moment I tested Fabel for the first time.
[06:46] And that's where I really see the return of investment when it looks through complex code bases, but also knowledge bases.
[06:55] If you watch my previous video where I talked about Sonnet and challenged Sonnet versus Opus, my local folder with my personal knowledge management in there has 168 GB and 160,000
[07:08] files in this folder.
[07:11] So, using a model that can really understand things holistically, you find out things that you wouldn't have catch previously.
[07:17] Would I use it just for my knowledge base?
[07:19] Not sure, especially if I need to pay per API.
[07:21] But when it comes to the business and optimizing systems inside the business, certainly I would do this.
[07:29] Here I show you the example of our iCodeJourney courses, but I worked in corporate in IT with engineering.
[07:37] I worked also as a validation expert with big data analysis in these corporates, and this would be insane if I would have had access to such a model to move forward in analysis much quicker to find the things.
[07:50] It's not about replacing human being with AI.
[07:52] It's about empowering the experts to get faster to the end result which saves essentially time for these experts and we can move faster.
[08:02] And here we need to keep in mind I might have got to a similar result with Opus through addition several
[08:09] iterations.
[08:11] But this is where the return of investment comes in.
[08:13] If I need to have run Opus several times until get I get a proper result versus I could use fable with a one prompt and I get the result that I'm looking for then the token payments are much less when I use fable obviously.
[08:25] So this is the result that's what it came up with but here I asked it specifically to also visualize the connections and that's the thing that I have here.
[08:34] So it visualized this using react flow so I'm not going into the details but it can do this out of the box.
[08:40] It's all based on my design system that the team already knows and if you follow this channel you know it's all based on a local folder.
[08:48] There's nothing special.
[08:50] It creates these HTMLs out of the box.
[08:52] If you want to get this too you can join us for free and download the scaffold for yourself and start building these things too.
[08:57] But here's the end result so we see when we zoom in here we see now here course 1 2 3 we see the cross connection and that's our whole mess.
[09:05] It's nice always to show off these things but it doesn't tell you
[09:09] anything.
[09:11] And that's why there is already a filter built in.
[09:14] It's a static HTML page.
[09:16] It's not an app but I can still make more sense out of the results by just say remove these connections and now I just see the dependencies.
[09:23] I can now pick just one course and now I see how this is connected or how this course is connecting to other lessons and go to another course and there you see the forward and the back connection.
[09:34] Here's the different things that's confusing.
[09:36] That's a minor issue.
[09:38] That's the blocking thing.
[09:40] So, I really can dive deep into this and here is the video expansions that it was recommended and there's a legend and so on.
[09:47] I can look for best case.
[09:49] This is really amazing if you I want to visually get my head around all the different lessons and how it is connected.
[09:57] To me, it's amazing to have this here, but then it dives in.
[09:59] I see the golden thread.
[10:01] That's what we are going for me to optimize our iCode journey courses that there is a golden thread going through the whole thing because this is what it is.
[10:07] Each course
[10:10] builds on top of the other.
[10:12] So, you really go through a productivity system end to end and that's where we start with note-taking and so on and you see between note-taking PKM there is already an intact golden thread, but then it disconnects because I said obviously in task management we teach how you set up routines, how you prioritize your day and your week and how you do weekly planning and all these things.
[10:32] So, obviously this needs to be when I want to get the task done backlinked to the information that I need to get the data task done and that's where this was for me the big moment where Faber will help us now to cross-connect even better the different courses that we have here and the different concepts and workflows with the previous lessons.
[10:50] And that's what he's now recommending here.
[10:52] And by the way, that's exactly what we will roll out today.
[10:56] The moment this resets because look at this, the sessions that I did, it burned through the tokens and Faber is already at 50% and resets in well, in over 2 days.
[11:09] Um you can only use Faber for 50% of the
[11:11] weekly usage.
[11:15] So, it will hard stop here and you see just the current session filled up within 1 hour.
[11:19] So, we already down the road, but when I finished the work, it nearly finished the full 100%.
[11:23] So, that's why I decided to stop here, make the video and show you my work in progress.
[11:27] But this is it, okay?
[11:29] I can go here.
[11:31] It shows me the course.
[11:33] I can open this up.
[11:34] It shows me the details about the course where I drifted, where it's in sync, what I need to look at, and so on.
[11:38] And obviously, I could do this manually, but look of the insane amount of work that I could hand over to a human team now to do this or developers.
[11:46] And this is where the return of investment sits now.
[11:49] If I just give a prompt and say, "Okay, roll this out."
[11:53] And I can trust the model that it gets, let's say, 80% there, that's more than enough.
[11:58] And I get this within 1 to hours.
[12:01] This is insane if you think about the people that need to work on this in order to get this done.
[12:06] And these people are now relieved because they can focus on other things again.
[12:10] In our case, it's just Paco and me now.
[12:10] We have no more
[12:13] any people working in my eye core to prove the concept that we don't need additional people inside the business to make it work, which doesn't mean that we are not working with people anymore.
[12:20] As I said, for Paco with his marketing company of 70 employees working in there and hundreds of vendors and so on involved, they got empowered with AI, too.
[12:31] The designers, they get AI access because prototyping is so much faster now.
[12:36] They get much faster going into the flow and get inspiration what to do.
[12:39] And then with their expertise, they can build much quicker something than starting everything from scratch.
[12:47] And that's, I think, is really underestimated what you can help your employees to empower with AI.
[12:51] So, you can see, we can go through the whole thing.
[12:55] There's a lot of details that came out here.
[12:57] And the moment my cap resets, I will hit enter, and this will get updated, no doubt about this.
[13:05] And this to me was proof.
[13:08] This is not a fable thing to make this report.
[13:10] As I said, this is usually the way that I want to get the reports back.
[13:12] I don't
[13:15] want to scroll through chat windows or terminals.
[13:19] How our folder structure works is that this ends up in these deliverables folders.
[13:23] So, when I work with sessions, they all for each session or topic create an individual deliverables folder, so I can always go back like yesterday, you know, when I published the video, here's the work in progress of the blog version and all these things.
[13:37] So, I can go in here, and once the things are done, because the team has an internal task management system, so they know if this work is done or not, they will end up archiving it the next time I close a session.
[13:46] Just as a side note, this is the efficiency of the team independent from the model, because all this and also the report creation, the design and output and all this has nothing to do with Claude or Fable or whatnot.
[14:01] I could get the same results with Codex, Gemini, GLM, it doesn't matter, because it's all based on instructions that I have inside the folder for the team.
[14:09] But, the comprehensive looking, the details in the lookings, this is where Fable really shines, and this is something you see
[14:16] the moment you compare it with Opus.
[14:19] And that's what I did here, but before I go in this, I want to show you this session that runs on Opus, okay?
[14:25] This is I then just used Opus to optimize the prompt, so I get a much better prompt up front for Fable to work with.
[14:34] Usually, this is what Larry is doing anyway.
[14:36] Larry is my orchestrator in this folder, so he knows how to interpret what I'm asking for and then hands it over to his sub-agents who are living inside this team here.
[14:44] But, in this case, I wanted to get if I'm already using Fable and I only have this small window to use it, I wanted to get it perfect.
[14:53] I mean, you already saw that in this one, even with this kind of prompt, I got an amazing output.
[14:57] But, for the other things, that's another thing I wanted to always do is improving the responsiveness of our web application.
[15:05] It's already okay, but to me, I want to optimize this perfect.
[15:10] I understand there are some scattered code snippets and things like that that we need to refactor and so on, and that's
[15:16] why I gave it another go for Fable to look at the front end performance responsiveness, and that's what we get here.
[15:23] This is now here I copy paste this prompt in.
[15:25] Faber went through this in no time.
[15:28] I mean, no time means 1 hour.
[15:31] And we get here the result again with details that I would have never received from Opus.
[15:36] And again, I will show you in a moment if we directly compare Faber with Opus results.
[15:39] But here you see it went deep.
[15:42] It it understands everything inside the code base what we need to look for.
[15:47] And this is all many things that I knew already, but it's such a comprehensive report that I now can just say, "Let's go.
[15:51] Let's do this."
[15:53] And I can be sure that the responsiveness will improve and everything will get ironed out because, and that's why I'm so sure, because I'm working with Opus since day one and I see the perceived quality improvements in everything that I tried to fix.
[16:11] And getting already such a report up front gives me enough confidence to then start rolling this
[16:18] out and let it just work.
[16:20] I really hope that I can do this within the token window.
[16:22] I'm I'm confident.
[16:24] But you see these things.
[16:25] That's why I love these reports.
[16:28] It even brings up mockups and explains to me what we will do and what we will change and so on.
[16:32] That's what I in particular ask for, but this is this is amazing.
[16:37] I love this type of feedback getting back from AI.
[16:39] And here we go.
[16:40] This is the example of improving responsiveness.
[16:42] And if you're already a member, you can tell me if you next week see some improvements too.
[16:47] Although, as I say, we could perfectly live with the quality that we already have here because everything is very fast loading and working already inside the membership.
[16:56] But we are constantly refining and this is something that's really worth because the better the foundation, the better we can add more features on top of it.
[17:04] And now we come into the comparison.
[17:06] That was the really interesting thing that I did here.
[17:08] So you see there are these two tabs fix password reset.
[17:13] And that was just something, a minor thing, that when people log in with Google authentication and they want to password reset, they
[17:18] get confused because it doesn't work because they logged in with Google instead of setting up a password in the first place.
[17:24] It doesn't matter about the details.
[17:28] It's just a minor thing that I wanted to check and to double-check if there's something that I might have missed.
[17:33] And this is where it really showed that it works.
[17:35] Now, I show you exactly how this worked.
[17:39] Here, I just said, "We need to fix password reset errors. Can you identify what the issue is?"
[17:45] And I handed over a ClickUp link that shows the task inside ClickUp where Paco shared some issues that he found with this particular, but I could also hand over link to our support ticket system or whatever because my folder here, the team working in here, they perfectly understand that this is a ClickUp link.
[18:02] We have a connection via API to ClickUp, so they have a connection.
[18:09] They can look up what is the content, and then they start working.
[18:13] As you can see here, Faber went through this in no time.
[18:15] It just took 8 minutes to figure out what the issue is and how to solve it.
[18:18] 8 minutes, all right?
[18:18] So, the thing is, I
[18:20] did the same with Opus.
[18:22] Exact same prompt.
[18:24] It went through the whole thing, and it took 3 minutes.
[18:27] Okay?
[18:27] It's much faster with Opus.
[18:29] However, what I then did, I pointed each other session to the other session.
[18:34] So, what does this mean?
[18:36] For those who don't know yet, if you use Claude in the terminal here, each chat session gets saved locally on your machine, the full chat conversation.
[18:43] So, even if you close the terminal or the system breaks down, the Claude conversation is still there.
[18:49] That's what you get when you hit resume here, and then you can choose the previous sessions that you that to pick up from.
[18:56] That's how it works.
[18:56] It's locally saved.
[18:58] It goes back into the session, and then it looks it up.
[19:02] You can show the session ID in the status line.
[19:07] If you just hit status line, and then write down, "Show me the Claude ID."
[19:10] Okay?
[19:12] And then it will set it up for you this way.
[19:14] You can tell whatever you like to have in the status line and that's something I have here.
[19:18] So all I did is now I picked up this ID from this session which was
[19:20] Opus, went to Fable and said look into this cloud session.
[19:23] What do you think about their findings?
[19:25] And then the other way around, look into the Fable session and compare the findings.
[19:29] And that's where the big difference comes in because I already noticed that Fable looked much deeper and found a much more elegant way to overcome this problem that we have here with the password reset compared to Opus which just recommended just to make a UI information that you have to log in via Google by versus here there's an actual proper solution.
[19:52] And that's the big difference that they figured both out.
[19:54] It says here, yes, the other session figured out the similar issues.
[19:56] However, their claim is wrong because they concluded the recovery flow is a dead end by design because there is no password reset.
[20:07] And that's not true because Fable dig deeper and they figured out and if Fable figured out there's a lot more we can do.
[20:15] And this is what then Opus confirmed that they agreed with
[20:21] high confidence but their but their run
[20:24] is better than ours. So he's
[20:26] acknowledging here that it was better.
[20:28] They went one layer deeper and it paid
[20:31] off and that says it all to me that
[20:33] Fable digs deeper into the things and
[20:36] gathers a lot more connections and
[20:39] interpretations than Opus does. And
[20:41] that's where he understood that there's
[20:43] a lot more and obviously now I could let
[20:45] a Opus do the implementation too but
[20:48] that's where I'm not so sure about. Just
[20:50] because I figured all these out and I
[20:52] get now these amazing reports, would I
[20:54] now use Opus to roll out for example the
[20:57] lesson upgrades and so on? Probably not.
[20:59] I will let Fable do it because I want to
[21:01] ensure while it's doing it to keep the
[21:04] same deep level understanding than while
[21:07] doing the report, that's just the trust
[21:09] that I have here and I could do some
[21:11] testing to see then also the outputs,
[21:13] but therefore it really becomes
[21:15] expensive when it comes to this cross
[21:16] testing obviously, because I run the
[21:18] same analysis with Opus in combination
[21:21] with Fabel and so on and this is
[21:23] something I might do in the future to
[21:25] then see is it worth saving money using
[21:28] Opus over Fabel for implementation and
[21:30] not only analysis and audit. And then
[21:32] you see here when we keep going, I made
[21:35] a lot of other prompts here too that I
[21:37] let optimize the things and I handed
[21:39] over to Fabel. So for example, there's a
[21:42] whole security audit that I ran on the
[21:45] web application and I'm happy that we
[21:47] were already on a great state and
[21:49] there's just some recommendations, but
[21:51] these are the things that I think Fabel
[21:54] is really useful and it's worth the
[21:56] money because you avoid wrong
[21:58] conclusions, that you go down a road
[22:01] that wasn't seen holistically. The thing
[22:03] is now obviously AI does mistakes, Fabel
[22:06] will also do mistakes. Is this the
[22:08] correct thing? Does it do everything
[22:10] correct with the lessons for example and
[22:12] so on? And that's where we really have
[22:14] to stop here and think about how is it
[22:16] to work with human beings? If I hire a
[22:19] person who is an expert in something and
[22:21] they need to dive in complex systems
[22:23] like this or even together as a team,
[22:25] it's also very likely that they come
[22:28] back with a conclusion that might be
[22:30] wrong. I worked in high stake projects
[22:33] where we talk about billions of budget
[22:35] and they were going things downhill due
[22:38] to some wrong conclusions and that's
[22:40] where it's always based on trying to
[22:43] gather as much information, compare
[22:45] information to each other as the
[22:46] decision maker and then make the
[22:48] decision based on the information that
[22:50] you gathered. And I think you can use AI
[22:53] really to get a starting point and let
[22:55] your human experts then evaluate or let
[22:58] them use the tools too, but then
[23:00] evaluate does this make sense? Maybe
[23:01] they oversaw something and that's where
[23:03] we can run three or four of the same
[23:06] agents doing the same work and then
[23:08] compare what they come back with because
[23:10] it will always vary. But imagine with
[23:12] humans you would do the same thing. If
[23:14] you have two people, experts in the same
[23:17] field, and you send them out to find a
[23:19] solution, it's likely that you will get
[23:21] different results back. And And that's
[23:23] where these people become bottlenecks if
[23:25] they're really good in identifying
[23:27] patterns and things like this and they
[23:29] suddenly leave the company or switch the
[23:31] team, then they have a big gap here
[23:33] where they try to backfill with a new
[23:35] person who has no clue what's going on
[23:37] and has his own experience with the
[23:39] things you use different systems and so
[23:42] on. So the quality of output varies
[23:45] drastically between different humans
[23:47] doing the same job. So here, I can buy
[23:51] this folder structure and the standard
[23:53] instructions that you have, something
[23:54] that you also do with humans by the way,
[23:56] with standard operating procedures,
[23:58] guidelines, and so on. So they follow
[24:00] something that you concluded in the
[24:01] past, more or less the same. And for AI,
[24:04] it's the same. Even that I have
[24:05] instructions, standard operating
[24:07] procedure, it's never following 100% the
[24:09] same thing because it's not an
[24:11] automation. But in this regard, it's a
[24:13] good thing because it thinks outside the
[24:15] box and uses the or interprets these
[24:19] instructions in different ways. And
[24:21] that's where I don't want to have an
[24:22] automation going through this and make
[24:24] analysis because then I cannot have
[24:26] these comparisons going on and seeing
[24:28] what they're looking up for. And auto
[24:30] mations and scripts still are relevant
[24:32] when you need to have proper data entry
[24:34] and so on. That's just something I
[24:36] wanted to add at the end. What about
[24:37] you? Do you use Fable already? Do you
[24:40] see already return of investment
[24:42] investing in such an expensive model?
[24:44] And let's face it, it is really
[24:46] expensive. So this could be thousands of
[24:49] dollars the analysis that I just did
[24:51] here. So is it really worth doing?
[24:53] That's where you need to understand the
[24:55] overall numbers in your business. Is it
[24:57] really worth when we optimize, for
[24:59] example, the courses, will it improve
[25:01] the user experience and the retention
[25:04] rate so they stay longer or is it more
[25:06] likely that they will consume the
[25:08] content end-to-end? That would be for
[25:09] this use case, but then it's the same
[25:12] for process optimization on a product
[25:14] assembly line, for example, where you
[25:16] need to understand is it really worth
[25:18] digging deep into this or not. But the
[25:20] same question you should always ask
[25:23] having human workers too. Is it really
[25:25] worth investing time that five people
[25:29] costing you $300 per hour sitting in a
[25:32] meeting talking about the same
[25:34] again that they already concluded 3
[25:36] weeks ago just because it's not clear
[25:38] where to store the information, retrieve
[25:40] it later, and so on. And that's why I
[25:42] core is so important because it teaches
[25:44] these fundamentals of single source of
[25:46] truth, where to store information, where
[25:48] to retrieve information, and all these
[25:50] things. So the human team can work much
[25:53] more efficient and focus on the things
[25:55] that matter instead of all this
[25:56] administrative friction that so many are
[25:58] suffering from. And once you have this
[26:01] and you put another layer on top like AI
[26:03] or automation, this is where everything
[26:05] really accelerates and you build
[26:07] momentum. And not the other way around
[26:08] to try to automate something that's
[26:10] broken because then you just amplify
[26:12] what's not working already and then you
[26:14] see these companies where they waste a
[26:16] lot of tokens on things that make no
[26:18] sense. And well, that's a whole other
[26:20] story. I think that's it for this video
[26:22] today. I hope it helped you to get some
[26:25] insights behind the scenes how we work
[26:27] inside my I core using AI, but also
[26:30] hopefully you can also extrapolate from
[26:32] this to your own work and see how you
[26:34] can take the most advantage of a fable
[26:37] in your use case instead of trying to
[26:39] build some games and see how far it can
[26:41] get. Here we have real use cases where
[26:44] it's really worth using AI. If you like
[26:46] the video, give it a thumbs up. And if
[26:47] you haven't already, subscribe to the
[26:49] channel so I can catch you up in the
[26:50] next one.