Going freelance as an AI PO: what I wish I'd known

I’m launching as a freelance AI Product Owner. Here’s what I’m learning as I live it — no filter, no manufactured success story.

The real question before launching

Not “am I legitimate enough?” — that question will paralyze you indefinitely. The real question is: for whom do I create value, and how?

For me, the answer is clear: tech teams with AI projects to launch but lacking someone who can bridge product vision and technical reality. My 10-year dev background is the argument.

Before launching, formulate your answer to this question. Not in terms of skills — in terms of problems solved for specific people.

The impostor syndrome, AI PO edition

It has a particular flavor in this field: “I haven’t shipped an AI project in production yet, am I legitimate?”

My answer: legitimacy doesn’t come from past projects, it comes from the value you bring today. If you understand LLMs, can write specs for agents, and have a technical background that lets you arbitrate architecture decisions — you have real value.

Be honest about your experience level. Clients looking for someone invested and rigorous appreciate honesty. Those who only want a portfolio of 10 shipped projects probably aren’t your ideal first clients.

How to find first clients

LinkedIn, seriously. Not to post vague content about “AI changing everything,” but to document what you’re concretely learning. A post on “here’s how I solved this agent spec problem” attracts the right people.

Your existing professional network. Your dev or PM network knows people with AI projects. Be clear about what you do now and for whom it’s useful.

AI communities. Anthropic Discord, LangChain Slack, AI meetups in your city. These are places where people have real problems to solve.

Short missions to start. An AI maturity audit (1-2 days), a spec review (a few hours), a scoping workshop — these are low-risk entry points for the client.

What I would have done differently

Document from day one. Every thing I learn, every problem I solve, every mistake I make — I write it now. That’s the blog. That’s my portfolio in real time.

Define my positioning earlier. I wasted time trying to be a “generalist AI PO.” My real advantage is specific: a PO who comes from dev. Owning that clearly from the start attracts the right clients.

Don’t wait to be “ready.” There’s no moment when you’ll be ready enough. You learn by doing. The first client is always the hardest to convince — but also often the one you learn the most from.