What is an AI Product Builder?
For the past week, I've been getting LinkedIn messages from people I don't know. They want to understand what an AI Product Builder actually does. How I got here. Whether we could chat for 30 minutes to "share our views on this new wild west." I've seen this movie before.
I've seen this before
In 2019, I pivoted to UX Writing. In France, nobody knew what it was. No job descriptions, no training programs, no community. When I said "UX Writer" in interviews, people asked if it was a type of graphic design.
Then the DMs started. "What exactly do you do?", "How did you get into this?", "Can we grab a coffee?" I met 4 or 5 people that way. Some became peers, others became competitors. The profession structured itself. Today there are jobs, training programs, a community.
Seven years later, same script. Word for word. Only the title has changed.
What it is
An AI Product Builder is someone who designs and builds products that use AI as a core component. Not someone who writes prompts. Not someone who makes slides about AI. Someone who ships.
The key word is "builder." You design, you code, you deploy. The product runs in production, real users interact with it.
It's also someone who operates across layers. You understand user needs (product), you can assemble the technical stack (dev), and you can shape the result (design). Not an expert in all three. Just enough to move forward without waiting.
What I build
To make this concrete: here's what I've shipped in the last 18 months, freelance:
Altaria. An AI literacy app for a real estate group. 145 activities, quizzes, games, Duolingo-style gamification. 3,000 users. Next.js, Zustand, Vercel AI SDK, Supabase. From idea to deployment, alone.
Fusil.paris. A full e-commerce platform for a jewelry brand. Catalog, Stripe and PayPal checkout, admin dashboard, transactional emails. In production, real customers placing real orders.
An automated accounting pipeline. Take a photo of an invoice with your iPhone, AI extracts the data, matches it with your bank statements. Telegram, Hono, Supabase, Revolut API.
A Figma plugin that extracts and analyzes UX content directly from design files. TypeScript, Figma Plugin API, multi-provider AI.
17 AI agents I use daily: copywriting, adversarial testing, parallel research, art direction, code auditing.
And Morphow, a modular SaaS webapp platform, currently in progress.
None of this was in a job description.
The profile
You don't need to be a senior developer. You need to know enough to ship. My background is wooden house construction, political communication, advertising, UX Writing, content design. And one day I decided to learn to code because I was tired of having ten ideas per screen and waiting for someone else to build them.
A UX or content background is an advantage. You think in user problems, not technical solutions. You instinctively ask "for whom?" before "how?" And you can write, which matters more than people think when half your product is AI-generated content that needs framing.
The real skill is knowing which AI capability to apply to which problem. And being comfortable with everything breaking every three months.
Why now
The tools are ready. Vercel AI SDK, Claude, Cursor, v0. A single freelance builder can ship what took a team of five two years ago. The barrier to entry has never been lower, and the speed of execution has never been higher.
Companies are starting to understand they need people who can think AND build with AI. Not just talk about it.
It's exactly like UX Writing in 2019. Nobody knows what it is yet. And that's why the DMs keep coming.
If you're one of the people who reached out this week: yes, this is what it looks like. And no, there's no playbook yet. That's the whole point.
