Big Idea

The argument in one line.

Now that any complex application can be built from two prompts in 45 minutes, the bottleneck in software has permanently shifted from execution to idea quality, distribution, and taste — the three things an AI model cannot generate for you.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…

An indie builder or founder who has been waiting for AI coding tools to reach production-quality output before committing to them.
Someone considering building a custom internal tool instead of paying recurring SaaS subscriptions.
A product designer or creative who wants to understand why taste and product instincts compound in value as execution costs collapse.
Anyone who wants an uncut, live demo of a high-capability agent building a complex multi-feature app from a single vague prompt.

SKIP IF…

You want a deep technical breakdown of architecture or how the AI reasoned through the build — the video stays at demo level throughout.
You already work daily with frontier AI coding agents and are past the can it do this stage.

TL;DR

The full version, fast.

Two loosely written prompts to Claude's latest model in Claude Code produced a fully functional Notion clone in 45 minutes: block editor, database views (board/gallery/list), emoji picker, light/dark mode, trash/restore, Convex backend — no spec, no PRD, no architecture guidance. The host's argument is that this flips the calculus for builders entirely: because execution is now cheap and fast, all leverage lives in choosing the right problem, knowing what good looks like, and having a distribution path. The practical recommendation: validate and spec with a cheaper model first, then hand the spec to the high-capability model to build the foundation in one session.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →

Chapters

Where the time goes.

00:00 – 01:25

01 · Introduction

Hook: Fable 5 / Mythos dropped; host tested it on a $10B app clone and the results shocked him.

01:25 – 03:30

02 · How to access Fable 5

Claude Code desktop app, model dropdown, note that subscription access ends June 22 after which API billing kicks in.

03:30 – 05:04

03 · The prompt

Host reveals the single vibe-coded prompt: macOS desktop app, Notion clone, Convex backend, no auth, no stack spec. Describes it as the most vibe-cody prompt possible.

05:04 – 06:12

04 · First look at the AI screenshot

Model spent 10 minutes thinking, planned the build, set up Electron, built the app, then self-screenshotted and self-corrected. Host says he cannot tell the difference from real Notion.

06:12 – 08:01

05 · Live demo — block editor

Host shows dark-mode app: sidebar, page hierarchy, inline block picker (lists/headings/tables/to-do), emoji picker identical to Notion, heading animations, cover image controls.

08:01 – 09:50

06 · Database and advanced features

Tasks database with board, gallery, and list views; per-item page view with tags/priority/status properties; filtering, sorting, new page creation inside a database.

09:50 – 11:04

07 · Light mode, settings, trash

Settings panel, light mode, typography options (serif/mono, small text, full width), trash/restore flow with bottom-left toast notification.

11:04 – 13:14

08 · The shift — what should I build?

Host pivots from demo to thesis: building is solved. The real question is what to build. Demand for software will explode; value moves to idea quality and distribution.

13:14 – 14:30

09 · The opportunity window

Your window is now — the gap between those who know this is possible and those who do not is today's opportunity. Distribution, taste, and idea quality are the three remaining moats.

14:30 – 15:45

10 · Cost and smart usage

Fable 5 is expensive; spec with a cheaper model first to validate the idea, then build with Fable 5. CTA to skool community.

Atomic Insights

Lines worth screenshotting.

A $10B app can be cloned to functional, visually accurate parity in 45 minutes with two loosely written prompts — no spec, no PRD, no architecture guidance.
The hard part of software is no longer building it; it is choosing the right thing to build and finding people who want to use it.
An AI that spends 10 minutes thinking before writing a single line of code produces dramatically better output than one that starts immediately.
Validating your idea and writing a spec with a cheap model before switching to an expensive one is now the single highest-ROI habit for AI builders.
Distribution, taste, and idea quality are the three things AI cannot generate for you — they are the new source of competitive advantage in software.
Building a personalized clone of an expensive SaaS tool is now a legitimate alternative to paying subscriptions, even for non-developers.
The window between people who know what AI agents can produce today and those who do not is the current opportunity — and it is closing.
A model that autonomously screenshots its own output, identifies failures, and self-corrects without a human prompt is a qualitatively different kind of coding assistant.
The model inferred Electron as the right architecture from a vague macOS desktop intent — stack specification is no longer a required input.
Design expertise and product taste compound in value as execution costs collapse — the rarer the judgment, the wider the gap between builders.

Takeaway

Validate the idea before you build the app.

WHAT TO LEARN

When a capable AI can build a production-quality complex app in 45 minutes from two sentences, the cost of building the wrong thing is no longer time — it is money and momentum.

Front-load idea validation: use a cheaper model to stress-test your concept before switching to a high-capability, high-cost model to execute the build.
Specifying a clear problem and audience before opening a build session is now the highest-leverage action available to a solo builder.
The three things AI cannot replace in software are distribution (finding users), taste (knowing what good feels like), and idea quality (picking the right problem).
A model that spends time planning before coding — researching the target product, laying out architecture, and self-testing its output — produces dramatically more complete results than one that starts immediately.
Building a personalized clone of a SaaS tool you currently pay for is a legitimate use of AI coding agents: you own the code, eliminate the subscription, and can extend it to your exact workflow.
The current opportunity window for builders is the gap between those who understand what AI agents can produce today and those who are still skeptical — that window closes as the demos become common knowledge.

Glossary

Terms worth knowing.

Fable 5 / Mythos: The video's branded name for Anthropic's latest Claude model. The host uses both names interchangeably throughout.
Claude Code: Anthropic's agentic coding interface that runs in the desktop app, giving the model filesystem access, terminal, and browser automation so it can build complete applications autonomously.
Convex: A real-time backend-as-a-service platform used here as the database layer for the Notion clone. Runs locally by default but can be pointed at a cloud project.
Electron: A framework for building cross-platform desktop apps using web technologies. The model chose Electron unprompted to fulfill the macOS desktop app requirement.
Vibe coding: Prompting an AI coding agent with loose, intent-driven language rather than precise technical specifications, relying on the model to infer the rest.
PRD: Product Requirements Document — a formal spec that defines features, scope, and behavior before engineering begins. The demo shows none was needed.

Resources

Things they pointed at.

01:25toolClaude Code (desktop app) ↗

03:30toolConvex ↗

15:15linkskool.com/aiapps community ↗

14:30linkreadytobuild.app ↗

Quotables

Lines you could clip.

05:50

“This has built this clone in just two simple prompts. No spec, no PRD, no roadmap, no complex setup, just these two prompts.”

Lands the demo headline claim in one sentence with zero context needed.→ TikTok hook↗ Tweet quote

11:26

“The question that you should be asking is not what can I do, it's what should I do?”

Contrarian pivot that reframes the AI coding narrative in nine words.→ IG reel cold open↗ Tweet quote

12:34

“Distribution, taste, design, and the quality of your idea — these are the three things that are key to building software that people will actually want to use.”

Clean enumeration of a clear thesis — quotable as a standalone slide or pull-quote.→ newsletter pull-quote↗ Tweet quote

The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

00:00Anthropic just dropped Fable five, which is their public version of the Mythos large language model, and this announcement is huge. And in this video, I wanna show you the power of Fable five in action. So I gave it the task of cloning an app that is worth $10,000,000,000 to see how well it would do and the results are absolutely crazy.

00:24I'm not even gonna like overhype this but it's kinda mad what it did. And so stick around if you wanna see the kind of crazy capability of this model. In this video, we're gonna break down how you can start using Fable five from Anthropic today, answer the question of can it clone a $10,000,000,000 app, and talk a little bit about what this means for the future of AI because this has huge implications on the capability of these AI tools in the future.

00:51If you don't know me, my name is Chris and for the last fifteen years have been designing apps and advising start ups on product and design. And if you are building apps and you wanna keep up with the latest in AI, then I have a community helping people do exactly that, where you'll learn not only how to build applications using AI, but also how to design them and how to get real customers.

01:10And as you'll see from Fable five, this is where the real skill comes into play. So without further ado, let's jump straight into the video and show you exactly what happened when I cloned a $10,000,000,000 app with Mythos a k a Fable five from Anthropy.

01:25So before I actually show you the app that Fable five created, I wanna show you we're just in Claude code in the desktop app at the moment. And I've got this set to Fable five max. And so if you do wanna start using Fable five, you can just go into the Claude desktop app or into the Claude website if you're just using it with chat.

01:41But if you wanna use it with Cowork and and Code, which is where I would recommend you start using Fable five because the capabilities are absolutely crazy, is to use it inside of Cowork and Code in the Claude desktop app and you're gonna have access to it through this drop down. Now, one thing to be be aware of is that access to Claude Fable five is only gonna be part of your Claude subscription up until June 22.

02:04After which, you're gonna be billed on an API cost which is gonna be so much more expensive than using us on your subscription. So use it now for building any projects that you wanna build. This is the time to get started with Fable five.

02:17This is the time to use it when you're using it against your subscription usage. This is the time to get the most out of it. But I don't wanna dig into Fable five and its capabilities and how you should or shouldn't use it too much because I'm gonna do a separate video on that.

02:29What I wanna show you here is whether it actually managed to clone this $10,000,000,000 app or not from this simple prompt. And the results here are are pretty crazy, but I'll let them speak for themselves.

02:42And so the prompts that I put in here, and this is literally the first test that I did with Fable five as well, is I added I sent this prompt off to Fable five to see what it could do. And I took the highest value app software company that I thought would be pretty difficult to build, which is Notion.

03:03So we're trying to recreate Notion here, basically, from scratch. And I thought for a lot of people, if you're working, you know, if you're working your own projects or if you're working in a business or you run a business, something that you might wanna create is your own version of Notion with a bunch customized tools inside of it so that you can do you can kind of create your own workspace for yourself without needing to pay for a Notion subscription.

03:26For example, you're kind of rebuilding your own software for your business or for your own projects. So what the prompt I sent off here is I want you to build a Mac OS desktop app. This should be an app that lets you create custom pages with tables, text, images, and more exactly like Notion.

03:41Use Convex for the database. No need for authentication as I'm just gonna use this for my own work. You decide the tech stack and implementation.

03:49The app should allow me to build my own custom workspace of pages, tables, and more just like Notion. So research Notion before you start building. Please build the full app.

03:58Make it incredible and a professionally designed product, and make sure everything works. This is like the most vibe cody of vibe cody prompts that you could possibly send without saying make no mistakes. But I'll be honest, this actually works with Fable five which is the crazy thing here.

04:15And so what Fable five did then was it spent a ton of time going off and doing research. It honestly spent about ten minutes just thinking before it even did anything. And this is one of the things to know about Fable five is that it does take a lot longer to do tasks.

04:29But it's gonna think things through really really in detail before it actually starts doing anything. So it's basically figured out the plan, figured out how it needs to build this, set the application up as an electron app, and also note here that we're building a desktop app. We're not we're not building this as a web app.

04:46So we're just creating our own desktop app that we can use on our own computer. It researched the brand, the editor modules, the database system, the different models that are needed inside of the product, and it set up the servers as well.

05:00It's marked the build phase as done. It's taken a screenshot here of the actual app that it's built. Now this is a screenshot of the app that it's built.

05:08I don't know about you but I don't think I could tell the difference really between this and actually Notion, which is crazy.

05:16That is insane. It says the menu looks right, grouped with icons and shortcut badges. It's gone and done a bunch of testing in the browser, taken screenshots, fed that back to itself, fixed any issues that came up.

05:26It's got light mode. It's got dark mode. And I'm gonna show you the actual application here because this chat thread is really really long for actually building the application.

05:34It built it fully, got the app up and running. There was one issue when it built it, which I had to prompt it for. So this has built this clone in just two simple prompts.

05:44No spec, no PRD, no roadmap, no complex setup, just these two prompts. I asked it to basically go through the app again and fix any issues you find including this one that I noted so that I can start to create pages and use the app. And it's gone back.

05:59It's fixed all of this and then given me an actual application that I can use that is a clone of Notion. So let me show you what this clone actually looks like and test it so that we can see if it works or not. Now this is the app that it created which is literally just Notion.

06:16Like, this is crazy. I don't understand how it's done this so well. Like, if you ask me, is this Notion or not?

06:22I would say, yes. This is Notion. It looks so close to exactly what Notion is.

06:28And not only is it just like a UI here, this is using the a convex back end. This is set up with an actual back end and database which is running on. Now, the database is running locally at the moment, but you could set this up as a as a convex project in the cloud just as easily.

06:44And I can not only, like, edit all of these things and go to this in the app, this inline control here, like, add in another element of all of this list of elements which is exactly the same as Notion, like a bulleted list table heading three to do list.

07:02We can add all of these things here. I can even highlight this and change the styling of this to a different text type. So we can change this to heading three.

07:12I can undo that. It's even got an animation for changing the the font sizes here. It's just crazy.

07:18I can check these off like, I don't I know this seems I know this seems trivial, but this is absolutely mad to me. The fact that we can just create this. It's even got change cover and remove on the cover for the page here.

07:32I can change this emoji. It gives me an emoji picker, the exact same as Notion. I can browse all the emoji categories.

07:39I can even search for an emoji. The level of detail here is just insane.

07:46Like, I don't wanna overhype this, but this is, a big deal when it comes to AI models and the capability. Now, obviously, this is incredibly expensive to run to be able to do this.

07:57But the fact that it's done this like, if I go to tasks here, we've got our different tasks. I can create different views of my tasks in this database, which is the same as a database in Notion.

08:10We've got all tasks. We've got a board. We've got a gallery set up.

08:13If I click on these, these are individual pages with these properties that I can add. With tags, priorities, status, I can add an icon or add a cover here, and this is gonna let me do that.

08:26And then that's got the icon there. I've got filtering. I've got sort by.

08:31I can add a new filter here for like priority. And it's got even all of these selections. We can search.

08:38I can click new to add a new item to this and then create this new page here which is under tasks. Even open this as a full page view. If I go back to tasks, I've got a reading list, is this view, meeting notes.

08:51I can add a new page, and I can even and I can add a new page from there. I can even go to settings here, change the settings in the mode.

09:01So we've got light mode here, which also looks incredibly good, and it's just Notion. It's mad. We've got all of these options in the top right here, so we can change it to serif or mono if we want to.

09:13Small text, so we can go smaller or full width. All of these options move to trash as well. If I move that to trash, you can see in you've got this even this pop up in the bottom left, which looks really good.

09:26I can go to trash. I've got these pages in trash. I can empty my trash or I can restore this back to my private collection of pages here within my Notion clone.

09:36I have not done any design tweaks or anything on this at all. This has just done this from these two prompts that weren't even that detailed with no spec or anything. I just think this is absolutely insane.

09:48Like, for anyone building applications, you now can build a complete application from one or two prompts that not just works, but is actually good and actually has a ton of functionality out of the box that is on par with a lot of the best software that exists in the world.

10:10I just this is I can't keep saying this enough, but it is absolutely crazy. It's unbelievable that you can do this. Um, I just think this is absolutely insane.

10:20If I do forward slash here, we can add in an image. I can upload my image, embed my image. All these blocks work.

10:26I can delete this. I mean, this is insane. This is absolutely crazy.

10:32Change the color of this. Delete this. Just I'll add this to favorites.

10:38It's gonna show up in favorites there. That's the exact same way that it works in Notion. I just honestly, add a database, add a subpage inside here, duplicate, move it to trash.

10:49I just it's honestly incredible the level of quality that you can get out of Fable five even with a simple prompt. So yeah.

11:00I wanna break down like my thoughts here on Fable five a little bit more because I think, like, now that Fable five can do this, there's a question in terms of how do you actually build things that people want? How do you build something that's good? And how do you build something that, you know, you could launch as a business or run as or run as part of your business, I think the demand for the amount of software that people want is gonna increase massively now that we can reach this level of quality in just a couple of prompts in building software.

11:31And I think the question that you should be asking if you're building applications, if you're building software is not what can I do, it's what should I do? What should you build? What is the right thing to build?

11:44What is the right idea for you to build? And what is something that can actually provide value for someone rather than or what is something that can actually provide value for you rather than just trying to go with whatever the opportunity is at any given moment in time.

12:00Like, turn your expertise into something that can help people actually achieve something. And the difficult part of building applications in software now is not the actual building part.

12:12That is pretty much done. We've got it here. You can use Fable five.

12:16If you had the money, if you can do this, you can clone Notion in and it did this in forty five minutes. That's the insane part of this. Two prompts, forty five minutes.

12:26The difficult part now is choosing the right thing to build, choosing the right idea, and then finding people who wanna use it. Distribution, taste, design, and the quality of your idea, these are the three things that are key to building applications and software that people will actually want to use.

12:45And this is also the exact things that I teach as well. Because my background is working with startup founders and as a designer leading design teams, and I'm so happy that you can build software this easily now because it means that there is so much more value for people still in learning these skills of finding the right idea, figuring out distribution, and designing a product that people will actually love to use.

13:10And with that said, let's get into my final thoughts on this. So as you can see, Fable five aka Mythos is absolutely insane when it comes to building incredibly complex, functional, working software even from just two prompts.

13:28And we've basically cloned an app that is a $10,000,000,000 business in two prompts and 45. And I'm not even just saying that.

13:37That's not hyperbole. It's literally what we did. And it looks great and everything works.

13:43Now we have a window now between the people who realize this is possible and the products that you can create and actually build and make money with.

13:53And so your opportunity is now. Your window is now. If you wanna build applications and software, if you wanna build products and you wanna actually make a difference in people's life with software, then now is the time to build it.

14:04Because this is a step change in what is possible with these AI models. And not even from a hype point of view, like the fact that I could do this is genuinely shocking to me, and I'm just amazed at what is possible to build with Fable five off the back of this.

14:22And I have so many ideas for products and I'm gonna build and take away and actually go and build off the back of this. But there is a couple of things to be aware of with Fable five, of course, if you are starting to use it and experiment with it on your Claude subscription. Number one is that it's incredibly expensive.

14:38And so it's now more important than ever to make sure you're really clear about building the right thing. So spend a lot of time, even with cheaper models like Opus, to really figure out exactly what you wanna build before you go away and build something.

14:52Because Fable five will be able to build a very complex, very complete product for you. It will cost a lot of money. But if you know exactly what you wanna build, and you validated that, you've got a good idea first, and that is something that I can help you with as well if you're interested in doing that over in my community.

15:06Come up with a solid spec with Opus before you go into building with Fable five and then let Fable five build that foundation of the product entirely for you in just a few prompts. And I didn't even use the goal feature here, and that is also what's so crazy about this. As I mentioned, if you are building with AI and you do wanna learn how to build a product that people actually want and get it into the hands of real paying customers, I do have a community for that over at score.com/aiapps.

15:31So you can click the link in the description down below to find out more and join the community there. Join a bunch of other people building with Fable five and experimenting with what it can do. If you enjoyed the video, don't forget to like and subscribe.

15:42Thank you for watching, and I will see you next time.

The Hook

The bait, then the rug-pull.

Two prompts. Forty-five minutes. A fully functional Notion clone with a working backend, database views, block editor, and emoji picker — indistinguishable from the real app. This is the demo that prompted the host to stop showing features and start asking a harder question.

Frameworks