A software MVP used to cost Brett three employees and 6 weeks.
A project manager to scope it. A designer to mock every screen in Figma. A developer to wire up the backend.
Now one person does the whole thing in under an hour. With Claude Code.
That's not a hypothetical. Brett Malinowski (The Brett Way, 500K+ subscribers) ran a software agency from 2022 to 2025 that did over $1 million a year building apps for creators. In his latest video he shows the exact model, then rebuilds the entire service delivery live, solo, in about an hour.
I pulled the whole thing apart. Here's the model, the numbers, and the stack.
The agency model: build apps for creators, not for random clients
Brett's agency had one positioning decision that made everything else work.
They didn't build software for whoever showed up. They built apps for creators with an audience.
Two ways to structure the deals:
- → A monthly subscription, usually around $6,000/month, to build the client's app
- → Or equity in the app, if they really believed in the idea
The monthly clients paid the bills. The equity clients were lottery tickets with good odds.
One of those tickets hit hard.
The equity deal that's now doing $500K/month
The agency took equity in Atlas AI, an app founded by Sebastian Ghiorghiu, an e-commerce YouTuber with over a million subscribers.
Atlas AI now makes over $500,000 per month.
Brett is no longer part of the day-to-day team and says he can't take credit for the growth. But he's clear on why the model worked, and it's not just the obvious "he had an audience" answer.
Sebastian had three unfair advantages:
- → He was his own customer. Years of e-commerce meant he deeply understood the problem, so his product instincts were sharp.
- → His network was massive. When you're a peer in a niche, people reply to you, want to work with you, and give you favorable rates.
- → The main growth channel wasn't even his own YouTube channel. It was the other channels he could sponsor.
That last one surprised me. The distribution advantage of being a creator isn't just your audience. It's your access to everyone else's.
Why this is an open opportunity right now
Here's Brett's pitch, and I think he's right.
Everyone is selling creators short-form editing, growth operating, or course funnels. Almost nobody is offering to build them software their audience would actually use.
A finance YouTuber needs a budgeting app. A video editing channel needs a VFX marketplace. A fitness creator needs a coaching tracker.
The DM is stupidly simple: "Hey, I'd be down to build you a custom VFX marketplace. I'd be happy to build it in exchange for equity. Worth a quick chat?"
And before you say "they'll never see my message": Brett has over 500K subscribers and gets maybe five DMs a day. He reads every single one. He's hired people from cold DMs.
Big creators see your DMs. They just don't need what most people are offering.
The three deal structures (and which one to pick)
If you're starting from zero, Brett keeps it simple:
- Monthly build fee. $2,000 to $6,000/month depending on your skill, the scope, and your proof.
- Equity. If you really believe in the creator or the idea, build for ownership.
- Hybrid. A smaller monthly fee plus upside in the app.
The hybrid is his favorite, and mine too. You're not broke while you build, but you still have exposure if the app takes off.
The creator brings the problem, the audience, and the distribution. You bring the software and the execution.
Now the interesting part: the execution used to require a team. It doesn't anymore.
The 1-hour MVP stack, phase by phase
Brett's agency delivered MVPs in three phases: PRD, design, then functionality. A three-person team, roughly 6 weeks.
Here's how he compressed each phase with AI.
Phase 1: The PRD writes itself
A PRD (product requirement doc) is just a clear list of what the app is, who it's for, and what v1 needs.
The old way: a project manager gets on the client call, takes notes, writes a long doc.
Brett's new way is a Zapier workflow:
- → Get on a call with the creator, record it with Fellow (an AI note-taker)
- → Zapier grabs the transcript and sends it to Claude with a system prompt ("you're a senior product manager, turn this discovery call into a build-ready PRD")
- → The finished PRD lands in a Google Drive folder named after the client
One pro tip from the video: bake your preferred tech stack into that system prompt. Then every PRD already tells Claude what backend, auth, payments, and hosting to use.
Client call ends, PRD appears. Zero human steps in between. If wiring up systems like this is new to you, this guide on business automations covers the same trigger-to-output logic applied across a whole service business.
Phase 2: Design (this is where most AI apps die)
Brett uploads the PRD to Claude Code and asks for a working localhost preview.
A few minutes later, he has a functioning marketplace. And it looks terrible.
His words: most AI-built apps look AI-built. That generic vibe-coded feel where the buttons work but nothing feels premium. If you're charging $6,000/month, it cannot feel like a toy.
The fix costs almost nothing:
- → A brand kit from Canva (search "brand guidelines" templates, customize, export the PDF)
- → A free component library from Figma (predesigned buttons, cards, navbars, modals)
Then one prompt: "Redesign the app using this brand kit and this component library."
A few iterations later the thing feels like a real product. And here's the compounding part: that brand kit and component library become your agency's signature style, reused on every future project.
Phase 3: Don't build hard infrastructure, plug it in
This is the smartest point in the video.
AI coding tools are great at v1s and bad at infrastructure. Try vibe-coding a real chat system. At first it feels simple, then real users show up and you need live delivery, notifications, file uploads, permissions, moderation. It gets ugly fast.
So don't ask AI to build infrastructure from zero. Ask it to plug in pre-built components.
For his VFX marketplace demo, Brett embedded Whop's pre-built components:
- → Payments: embedded checkout so sellers can list products, collect payment (card, crypto, Cash App), and withdraw to their bank
- → Chat: a buyer-to-seller DM system, added with one prompt in a few minutes
- → Auth: OAuth account creation, with an onboarding flow copied structurally from the best apps on Mobbin (screenshot the flow, tell Claude "build ours inspired by this structure")
Then Supabase for the database (users, listings, purchases, categories) and Vercel for hosting on a custom domain.
That's the entire MVP that used to take his agency 6 weeks.
The margin math that makes this scary
Brett's agency ran at roughly 55% profit margins. The main expense was headcount.
AI didn't just make the coding faster. It compressed the whole workflow: the PM's discovery call becomes an automated PRD, the designer's Figma work becomes a brand kit plus component library, the developer's infrastructure sprint becomes embedded components.
One person now delivers what took a team of three or four.
And because the stack is baked into every PRD, project two, three, and ten all ship with the same design system and the same infrastructure, automatically.
I've been saying this for months: AI services businesses are the fastest path to cash flow right now, and this creator-equity angle stacks long-term upside on top. It's the same systems thinking Brett laid out when he showed how to automate your entire job with AI, just pointed at service delivery instead of your own tasks.
What I'd steal from this if I were starting today
Not the VFX marketplace. The positioning.
- → Pick a niche you actually understand
- → Find educational creators in that niche with engaged audiences
- → Offer to build the app their audience is already asking for
- → Structure it hybrid: small monthly fee plus equity
- → Deliver with the AI stack above, in days instead of months
The build is no longer the moat, distribution is. And creators are sitting on distribution with no software team.
You be the software team.
FAQ
How much does it cost to build an app for a creator?
Almost nothing in tooling. Claude Code, Zapier, Canva, a free Figma component library, Supabase, and Vercel all have free or cheap tiers, plus a domain. Brett charged clients $2,000 to $6,000/month for this work, so the margin on your first deal is nearly all profit.
Do I need to be a developer to run this model?
No. Brett's whole demo is aimed at non-technical builders. Claude Code writes the code, and the hard infrastructure (payments, chat, auth) comes from pre-built components you embed instead of build. You need taste, clear communication with the client, and the discipline to follow the steps the AI gives you.
How do I get creators to respond to my DMs?
Keep it short and specific. Name the exact app you'd build for their audience and the deal you're offering. Brett gets around five DMs a day at 500K+ subscribers and reads all of them. Ignored DMs usually mean a vague offer, not an unseen message.
Is equity or a monthly fee better for a first deal?
Hybrid. Take a smaller monthly fee (even $1,000 to $2,000) so you're not building for free, plus a slice of equity for the upside. Pure equity is only worth it when the creator has serious distribution and you'd bet your own money on the idea.
Steal playbooks like this every week
I break down exactly how bootstrapped founders build to $100K/month and beyond on the Profitable Founder Podcast: real numbers, real deal structures, no fluff.
