Modern Creator
Luke Carter · YouTube

I Gave Claude Fable 5 One Prompt. It Built My Whole Backend

One prompt, a pre-written architecture repo, and a Claude-orchestrates-Codex workflow turn a static redesigned site into a fully autonomous CMS that writes, illustrates, and publishes its own articles.

Posted
today
Duration
Format
Demo
hype
Views
459
25 likes
Part of the collectionThe Fable 5 PlaybookAll 45 Fable 5 breakdowns, synthesized into one page.
Read the playbook
Big Idea

The argument in one line.

A single context-rich prompt let an orchestrator AI model delegate the heavy coding to a second execution model and build a fully working, autonomously-publishing CMS backend without the operator writing or touching any code.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…
  • Web designers or agency owners who want to sell backend/CMS systems to clients instead of one-off static websites.
  • Solo operators already using an AI coding assistant who want to hand off backend plumbing (database, CMS, publishing pipeline) rather than wiring it by hand.
  • Anyone maintaining a marketing site that's gone stale because there's no way to publish new content without a developer.
  • People curious about multi-model AI workflows where one model plans/strategizes and a second model executes the grunt work.
SKIP IF…
  • You want a reproducible step-by-step tutorial with exact prompts shown — this is a narrated demo of a result, not a walkthrough.
  • You don't have access to both an orchestrator-tier AI coding subscription and a second execution tool, since the workflow depends on running both together.
TL;DR

The full version, fast.

A designer who'd already migrated his site off Framer had no way to publish content — no CMS, no backend. He fed Claude (used here as an orchestrator he calls Fable 5) a pre-written backend architecture repo and told it to delegate execution to Codex GPT-5.5, giving it business context, not just instructions. The result: a Supabase-backed CMS with a Kanban content pipeline, scheduled trend scans, and an autonomous loop that writes, illustrates, and publishes SEO-optimized articles in his brand voice, gated by a manual-approval 'safe mode.' The lesson: pairing a strategist model with an execution model, and handing it a concrete architecture spec instead of open-ended instructions, is what turned a vague goal into a working system 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.

00:0000:45

01 · Autonomous CMS Setup with Claude AI

Cold open: the front end was already migrated off Framer in a prior video, but with no CMS there's no way to publish. Today's goal is to get Claude to build the backend, then later a CRM.

00:4501:18

02 · Digital Sovereignty & the Digital Home Explained

Introduces 'digital sovereignty' — owning your own front end, back end, data, and analytics instead of renting platforms that dictate your workflow.

01:1803:17

03 · Claude Fable 5 + Codex Orchestration Workflow

Switches to Fable 5, shown burning through weekly usage credits. Explains the workflow: Fable acts as strategist/orchestrator, delegating heavy execution to Codex GPT-5.5. Notes Anthropic's own guidance to explain 'why' when prompting, not just 'what.'

03:1704:10

04 · Installing the Digital Home Backend Repo

Pastes a prompt describing the desired autonomous publishing workflow, then pulls up the open-source 'Digital Home Backend Starter' GitHub repo and hands it to Claude with instructions to review and install it into the existing front end.

04:1006:51

05 · Why Web Designers Should Sell Backend Systems

Argues pretty front ends are commodified now that AI can replicate them in one prompt; the resellable value is in backend systems (content, leads, data) — pitches his Skool community that teaches the full setup process.

06:5108:05

06 · AI Publishes First Article Live (Untouched)

Returns to find the autonomous publishing loop verified end to end — a full SEO/GEO-optimized article with a generated hero image and FAQ block already live on the site, written in his brand voice pulling from real client results.

08:0509:35

07 · Inside the CMS Kanban Backend

Tours the new backend: a Kanban-style content pipeline where Claude has already planned multiple articles, with the live published piece visible and editable.

09:3510:37

08 · How the Autonomous Publishing Loop Works

Explains the mechanics: a built-in content-writing skill inside the backend repo lets Claude's API write articles in brand voice and auto-publish to the front end; only an Anthropic key and an OpenAI key (for images) were needed manually.

10:3711:30

09 · Draft to Published: Frontend Auto-Publishing

Shows a drafted article moving through review and appearing live and fully formatted on the front end, powered by the shared Supabase database between front and back end.

11:3012:42

10 · Next: Building an AI-Native CRM

Recaps the fully working autonomous CMS and teases the next video: building a custom CRM (leads, emails) into the same backend, designed for AI agents rather than clunky legacy CRM tools.

Atomic Insights

Lines worth screenshotting.

  • A pre-written backend architecture spec turned an open-ended 'build me a CMS' request into a concrete, buildable contract for an AI agent.
  • Splitting AI work between a strategist model and an execution model conserves the pricier model's usage budget for decisions, not grunt work.
  • Telling an AI assistant why you want something, not just what to build, is guidance the model's own maker gives because added context changes output quality.
  • A fully autonomous content system only counts as proven once it completes write, illustrate, review, and publish without a human touching any step.
  • Front-end web design is increasingly commodified because AI can replicate a polished site's look and feel in a single prompt.
  • The resellable skill for designers is shifting from pixel-pushing front ends to building the backend systems — content, leads, data — that clients actually depend on.
  • A shared database between a site's front end and back end is what makes publishing instant instead of requiring a separate deployment step.
  • A Kanban-style pipeline with a manual approval gate lets an otherwise fully automated content system still get human review before anything goes live.
  • A recurring scheduled trend scan is what prevents an automated content calendar from running out of relevant topics to write about.
  • Once a backend is built, extending it with a new function like a CRM costs less than starting a separate system because the database and hosting already exist.
  • Heavy reliance on one AI subscription's weekly usage allowance is a real operational constraint, not just a minor annoyance, when a daily workflow depends on it.
Takeaway

Orchestrating two AI models can build what one alone can't afford to.

AI ORCHESTRATION

Pairing a strategist model with an execution model, and handing it a concrete architecture spec instead of open-ended instructions, turned a vague 'build me a CMS' goal into a working autonomous publishing system in one sitting.

01Autonomous CMS Setup with Claude AI
  • A CMS is the bottleneck that stops any redesigned site from generating new content, no matter how polished the front end looks.
  • Framing an AI request around a business goal, not just a technical spec, gives an agent enough context to make good architectural calls on its own.
02Digital Sovereignty & the Digital Home Explained
  • Owning your codebase, data, and hosting instead of renting SaaS tools removes platform risk that compounds as your business grows.
  • A business built on infrastructure it doesn't control is exposed to pricing changes, feature removal, or shutdown decisions made by someone else.
03Claude Fable 5 + Codex Orchestration Workflow
  • Splitting AI work between a strategist model and an execution model conserves the expensive model's usage for judgment calls, not grunt work.
  • Telling an AI assistant why you're doing something, not just what to do, is guidance the model's own maker gives because added context measurably improves output quality.
  • Heavy reliance on a single subscription's usage allowance is a real constraint worth planning around before committing a daily workflow to it.
04Installing the Digital Home Backend Repo
  • Handing an agent a pre-written architecture document gives it a concrete contract to build against instead of open-ended instructions.
  • Retrofitting a backend onto a front end that wasn't designed for it is inherently backwards, but a detailed enough spec can still bridge the gap.
05Why Web Designers Should Sell Backend Systems
  • Front-end design alone is increasingly commodified because AI can replicate a polished site's look in a single prompt.
  • The differentiated, sellable skill shifts to building and owning the backend systems — content, leads, data — that a front end merely displays.
  • Packaging a repeatable backend system as a resellable service adds a recurring income stream beyond one-off design work.
06AI Publishes First Article Live (Untouched)
  • A system is only proven autonomous when it completes a full cycle — write, illustrate, review, publish — without a human intervening at any step.
  • Brand-voice consistency in AI-generated content depends on feeding the system real reference material, not generic instructions.
07Inside the CMS Kanban Backend
  • A Kanban-style content pipeline gives a human just enough manual checkpoints to stay in control of an otherwise automated system.
  • Requiring manual review before publishing is a reasonable default when first trusting an autonomous content system with a live site.
08How the Autonomous Publishing Loop Works
  • A recurring scheduled scan of search and news data is what keeps an automated content calendar from running out of relevant ideas.
  • The only ongoing manual inputs an autonomous publishing system needs are API keys and periodic approval clicks — everything else can run on a schedule.
09Draft to Published: Frontend Auto-Publishing
  • A shared database between the backend and the front end is what makes 'draft to live' instant instead of requiring a separate deploy step.
  • Reviewing a finished draft before publish is a fast way to catch tone or factual mismatches before they reach a real audience.
10Next: Building an AI-Native CRM
  • Off-the-shelf CRM tools are frequently criticized as clunky specifically because they weren't designed for AI agents to operate inside of.
  • Extending an existing backend to add a new function is cheaper than starting a new system, because the data layer and hosting are already solved.
Glossary

Terms worth knowing.

Digital Home
The creator's term for a business's self-owned front end and back end pair — a public website plus the content, lead, and data systems behind it — hosted on infrastructure the business controls instead of rented SaaS tools.
Fable 5
The AI coding assistant used as the 'strategist' in this workflow, shown on screen with its own weekly usage-limit dashboard; the creator pairs it with Codex for execution.
Codex GPT-5.5
The AI coding tool used as the 'executor' in this workflow, doing the bulk of file edits and setup work while the strategist model plans and reviews.
GEO (Generative Engine Optimization)
Optimizing content so it gets surfaced and cited by AI search tools and chat assistants, not just ranked in traditional Google search.
Kanban CMS
A content management interface organized as columns (e.g. planned, approved, writing, draft, published) that a human or AI agent moves content cards through.
Safe mode
A setting requiring manual review and approval before an AI-drafted article publishes automatically, used as a guardrail while trusting a new autonomous system.
Orchestrator / executor pattern
A workflow where a strategist AI model handles planning and judgment calls while delegating the bulk of implementation work to a second, execution-focused AI model.
Resources

Things they pointed at.

00:00toolWispr Flow
01:40linkAnthropic's prompting guidance (explain 'why', not just 'what')
03:23linkDigital Home Backend Starter (GitHub repo)
Quotables

Lines you could clip.

06:59
Your entire autonomous publishing loop is verified and live end to end.
the payoff line of the whole demoTikTok hook↗ Tweet quote
07:05
I literally didn't touch anything, and it's actually already gone and published an article live on our page following our exact brand voice.
clear before/after claim with a concrete resultIG reel cold open↗ Tweet quote
06:33
Pretty websites are completely commodified right now.
punchy contrarian claim about the design industrynewsletter pull-quote↗ Tweet quote
01:30
I'm burning through Fable credits like it ain't nobody's business.
honest cost admission that undercuts the hypeTikTok hook↗ 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.

metaphoranalogy
00:00Alright, guys. In this video, we're gonna be taking our front end designed website, and we're gonna be connecting an automated CMS system into the back of it so we can focus on automated publishing and get some articles moving again because, ultimately, this has been set stale for six months plus just gathering dust on Framer.
00:18And in my last video, I basically one shot transferred this over to my own code base. But the problem is I don't have any back end. I don't have a CMS system.
00:26I have no way of actually publishing new content, so we're kinda stuck. In this video, what I wanna do is put Claude and AI to the test and see if we can one shot a beautiful CMS system so we can go into fully autonomous content publishing, and then we can move into actually generating and building a CRM system, which is also completely vibe coded.
00:45So for those of you that haven't been following along, what we do at Brave Brand is we build digital homes for our clients, which are essentially digitally sovereign back ends and front ends so you and your clients can completely control and own your code base, your data, your analytics, everything, and not be reliant on any external subscriptions or platforms who ultimately dictate what you're gonna be doing.
01:07Alright? So this is all about a push and a play to digital sovereignty. In the last video, we were escaping the confines of Framer.
01:13And in today's video, we're gonna be focusing on building the back end and the system around this website. Alright. So what I'm gonna do first and foremost is I'm gonna go over to Claude, and this is in the same project that I set, uh, the Braebrand website up in.
01:26So what I'm gonna do now is just switch to Fable. You can see I've been absolutely wrecking my, uh, Fable credits.
01:33Alright? So, you know, this thing spends and burns credits way too fast in my opinion, and it's just, you know, it's really quite frustrating even though it's been incredibly powerful. So what I've actually done is I've created a workflow and a specific skill that allows Fable to be the strategist and the orchestrator, and it calls on Codex GBT 5.5 to do a lot of the heavy lifting.
01:54So, hopefully, we'll be able to get this done within our 10% credit for the rest of the week because we're only on flipping Wednesday, and that's about five days, you know, we're gonna be hanging dry for. One thing as well, you know, and this is through Anthropic's own guidance, is that when you are prompting Fable, and I found this quite interesting, is that they specifically say and encourage you to tell Fable why you are doing something.
02:18Because when you give it that extra context, apparently, it makes a difference. So that's what we're gonna try today.
02:23Okay? We recently transferred our bravebrand.com site to our own code base, which is now hosted on Cloudflare.
02:29In today's session, I wanna focus on getting our CMS system set up so we can start auto generating and auto publishing content directly to the front end of our site that is SEO and GEO optimized so that we can try and improve the organic inbound leads that we are getting from AI search and organic Google search. I'm gonna give you a specific back end repo, which details the plumbing that I would like you to follow because we have a very specific content writing and content publishing workflow baked in this repository.
03:01So I want you to review it, and I want you to install it into this front end so we can orchestrate and publish content completely autonomously through the back end into the front end and increase our organic traffic through Google and GEO. Okay. So that prompt is in.
03:16I've given it a little bit of why. And so what I'm gonna do now is just go over to my GitHub, where you'll see I've got a digital home back end repository here.
03:25Now for those of you that don't understand what a digital home is, essentially, a digital home is your front end, which is the public facing website, and your back end, which contains your CMS system and your CRM system. We've already built this entire infrastructure and all of the plumbing, which says what the database is gonna be, etcetera, etcetera.
03:42So the problem is we're doing this kind of backwards because the front end's already created, but it's not been created with this in mind. So I'm gonna take the back end, and I'm gonna copy that. And I'm just gonna paste this over into Claude, and I'm gonna give it the context.
03:56Alright? And I'm also just gonna say here, please use the codex workflow for this. Okay?
04:02Which essentially means that Fable is gonna do the thinking, the strategy behind this, and it's gonna orchestrate codex to do a lot of the heavy grunt work. Now in regards to this digital home setup, this is exactly the process that we take people through step by step inside of our school community. Alright?
04:18So if you're building new sites from scratch or if you wanna learn how to sell this as a service to your clients, then the school community is the best place for you to start because we've got a step by step process that guides you through every step of the way. So if you're a web designer who is used to designing pixel pushing, etcetera, etcetera, learning how to build this system back end, front end is going to increase your value in the marketplace because you're gonna be getting a plug and play system that you can ultimately resell to other people and add another income stream into your business.
04:47Because I've said this before, and I'll say it again, designers are incredibly well placed in this new industry. I really believe that designers are actually more important than ever before, but only if you can start learning how to market to and sell the outcomes that your clients are gonna be getting rather than just creating pretty websites because pretty websites are completely commodified right now.
05:09So the digital home thesis is that your clients and yourself are going to own the land you stand on. It's kinda like investing in land ten, twenty years ago, and the value of that is only going to increase as AI gets smarter and as the Internet grows and grows and grows. So if you can build your digital home on land you own right now, which is essentially owning your own code base, your own data, etcetera, you're well on the way to being digitally sovereign.
05:35Right? So these are open source repos, which you can basically get access to now, but the entire process and guidance on how to set this up properly for yourself and clients is inside of our school community. Now I'm kind of going on a whim here, and I'm assuming that Fable is smart enough to just pick this up without actually seeing the front end starter because this website has got nothing to do with that.
05:56Alright? This is just a pretty looking website that was created on Framer and is now been transferred to my own code base because Fable replicated it in one flipping prompt. Alright?
06:07So if you wanna see that video, here it is. But this is the follow on from that, and I'm hoping that Fable is going to be able to understand the process and execute it, you know, with a little bit of help from Codex. Alright?
06:18So let's let this work, and we'll see how it gets on. So if you wanna learn more on how to use Fable as an adviser, I recommend downloading this skill from Pete over here. Alright?
06:28So you literally just copy and paste this URL, drop it into Fable, and it will install the skill, which will allow Fable to orchestrate and then get Codex to do all of the hard work. Now, obviously, you'll need Codex authenticated on your machine, slightly more advanced workload, but, honestly, it saves credits and makes your work a lot more efficient because, like I said, I'm burning through Fable credits like it ain't nobody's business.
06:51Alright, guys. So has been cooking away, and it's been doing some really interesting stuff. But it's basically just said, this is kind of freaky, is your entire autonomous publishing loop is verified and live end to end.
07:03Alright? So I literally didn't touch anything, and it's actually already gone and published an article live on our page following our exact brand voice. It's generated a hero image, questions block, SEO, GEO, service confirmed, FAQ page, blah blah blah, all up and running.
07:20Okay? Um, and it's also created our back end. So let's first and foremost take a look at our articles page live on our site, and you'll see there that we've got an article published.
07:29Why are marketing, uh, keeps flopping without a brand strategy? Image generated end to end. Article written on a tone of voice, see client results just like this in detail.
07:40So it's like pulling from our actual clients. Oh my goodness. Which is insane.
07:44Alright. So that's all coming from our brand's brain, which we've already created, by the way.
07:49So this now links to our case studies. We've got a CTA directly to our school community. Holy crap.
07:57Alright? FAQs, call to action.
07:59Alright? So that's boshed and done and just implied because of the back end that we've set up.
08:05Now let's take a look at our actual back end for this. So our entire digital home back end has been set up end to end. Alright?
08:13So you can see here that the back end is essentially our CMS system where we've got this Kanban board view, and Claude's gone and planned a bunch of content already. And you can see the published article that is literally on our front end right now is here. So I can click into this.
08:27I can edit the entire thing. I can make changes. You know, I can also unpublish this from my site, but I can see it live, and everything works end to end.
08:37Alright? So let's go for the moment of truth now. So let's see if this will start writing this automatically.
08:43Alright. So this is essentially the test that happens in the back end of the digital home is you simply need to take ideas from here that are written through a skill that is inbuilt into the back end repository that I gave Claude code. So that entire back end repository, it has a bunch of content writing skills and content strategy skills baked into the back end already.
09:03So that's how Claude code actually went out and wrote these ideas. Alright? So when you push it into here, this now connects to Claude's API that writes the articles in my brand's tone of voice and publishes it directly to our front end.
09:16Alright? So that's the autonomous publishing loop, which Claude basically did.
09:20And all I had to do was add, uh, my Anthropic API key and my OpenAI API that generates the images, and then it gave me a link for my back end, which I just need to connect onto a proper URL. But the entire process, I mean, has literally been set up end to end. Alright.
09:36So for those of you that are not familiar with digital home, essentially, what has happened is we had a front end at the start of the video. So this is the public website that people see, and we literally dropped in our back end repository and gave that to Claude, and it connected this.
09:51It set up our entire back end. It set up a super base account, which is a shared database where all of our articles and content gets stored. So every time the back end writes an article to the super base, it publishes automatically onto our front end, and everything is hosted on CloudFare.
10:07And I didn't touch anything, guys. So the beauty of having a digital home is that your front end and your back end is optimized for AI agents to do the editing in the hard yards for you. A lot of people say to me is like, oh, but, you know, why don't I use Framer or WordPress, etcetera, etcetera?
10:22The reason being is I don't want to actually go around and drag and drop and click things anymore. I just want my agents to do it for me completely hands off. And that's what a digital home is ultimately all about, and you can see how quickly and easily that was set up.
10:36Alright. So you can see now this article has been drafted, and we can review it. Alright.
10:42Looks good. If we just click view, it shoots us over directly onto our front end, and we've got a perfectly formatted SEO optimized article done and dusted.
10:53Okay? So you can see that, you know, we're starting to create these topical clusters around our different brands pillars, and now we're back into autonomous publishing. Alright?
11:04So you can see just like that, you know, we dropped in the back end repo. Claude figured it out. It set up a super base account, uh, the shared database between these two sites, you know, and now we're ultimately rocking and rolling.
11:17Okay? So we've got this pipeline. I can just push these over here planned, and then the daily schedule, which is set up in GitHub actions, is just gonna write them and publish them because we're on autonomous mode.
11:29Alright? So that's fantastic. Now the next test that I wanna put Fable to is I wanted to create a fully customized CRM system built for the age of agents because, frankly, CRM systems today are clunky.
11:43They suck. If anytime I get a client into GHL, they go, oh my gosh. I don't know if I can do this.
11:49Alright? So we wanna create completely personalized CRM system, and we're gonna bake it into this back end.
11:54Alright? So it's gonna be baked in there with leads, emails.
11:58You know? I'm gonna let Fable figure out exactly what to do. So my friends, that's gonna be the next video.
12:02So if you'd like to see that, make sure you subscribe, stick around, and we'll share that very, very shortly. But you can see now we've got fully autonomous CMS system working, connected, writing articles to our front end, and now our site can start building up that topical authority and driving some organic leads through SEO and GEO.
12:21What a time to be alive, my friend. That is really what I mean by digital sovereignty. It's owning your data, owning your back end, owning your front end, your own code base, and letting agents do the heavy lifting.
12:32Alright? Because you have better shit to do, like chat to clients, get clients, make money, have fun.
12:38So, anyway, shut the hell up. I'll see you guys on the next one.
The Hook

The bait, then the rug-pull.

A rebuilt front end with nowhere to publish is just a pretty dead end. This video picks up right where that problem was left: a designer hands his AI coding assistant a single, context-loaded prompt and a pre-written backend spec, then watches it orchestrate a second AI model into building an entire CMS, database, and autonomous publishing loop before he's touched a line of code.

Frameworks

Named ideas worth stealing.

06:25model

Orchestrator / executor split (Codex-First pattern)

  1. Fable 5 (executor advisor) plans strategy and reviews output
  2. Codex GPT-5.5 (executor) performs the bulk of implementation work

A publicly shared skill (referenced via a GitHub link in the video) formalizes using one AI model as a strategist/advisor and a second as the execution workhorse, so the pricier or usage-capped model is only invoked for judgment calls.

Steal forany coding workflow where a usage-capped assistant needs to conserve its allowance while still directing a larger build
09:55model

Digital Home architecture

  1. Frontend (public website): homepage, blog, services, contact, SEO detection
  2. Backend (operating system): content pipeline, lead management, email, analytics, agent oversight
  3. Shared Supabase database connecting both
  4. Cloudflare hosting for the whole stack

A four-part architecture the creator uses to frame 'owning your stack': a public site and an operating-system-style backend, tied together by one shared database, hosted on infrastructure the business controls.

Steal forproductizing a backend-as-a-service offer for web design clients
CTA Breakdown

How they asked for the click.

VERBAL ASK
05:00product
the school community is the best place for you to start... learning how to build this system back end, front end is going to increase your value in the marketplace

Soft pitch woven directly into the technical value section rather than a hard sell — positions the paid Skool community as where to learn the reproducible process behind what's being demoed.

FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
OTHER LINKSAlso linked in the description.
Storyboard

Visual structure at a glance.

open
hookopen00:00
orchestration workflow
valueorchestration workflow01:18
sell backend systems
ctasell backend systems04:10
first article published
valuefirst article published06:51
publishing loop explained
valuepublishing loop explained09:35
CRM teaser / close
ctaCRM teaser / close11:30
Frame Gallery

Visual moments.

Watch next

More from this channel + related breakdowns.

47:51
Riley Brown · Interview

OpenAI Merges ChatGPT and Codex

Riley Brown and Ras Mic dig into GPT-5.6, Codex's background computer-use, and why self-scoring agent loops are turning coding tools into a general operating system.

July 12th
Chat about this