In March 2024, I made a decision that surprised many people in my professional circle: after 10 years of writing code, I decided to pivot to AI Product Management.
Here’s why, unfiltered.
The turning point: a project that opened my eyes
It all started with an internal project I worked on in late 2023. The goal seemed simple: automate incoming document processing with an LLM. In practice, it was a mess.
The project PM didn’t understand why “asking the AI to do X” didn’t produce X. Devs kept implementing specs that ignored the real constraints of LLMs. Meetings went in circles.
I was in the middle — a developer who understood both sides — spending more time translating, arbitrating, and reframing than actually coding.
That’s when I realized: the role that bridges product vision and the technical reality of AI was missing. And I was probably well-positioned to fill it.
What 10 years of dev gave me
When I say my technical background is an advantage, I’m not saying it to reassure myself. I say it because I’ve seen what it changes in practice:
I know what’s feasible. When a stakeholder comes with “we want an agent that does X,” I can answer immediately: feasible, feasible with these constraints, or not feasible at all.
I speak the same language as devs. No friction, no translation needed. When I say “the system prompt needs a JSON output schema to prevent hallucinations,” the team knows exactly what I mean.
I recognize false good ideas. AI is sold as a magic wand. With 10 years in production behind me, I can spot solutions that seem simple but will create technical debt for months.
The doubts I had
“Am I shooting myself in the foot by leaving a solid technical career?” Yes, I thought about it. My answer: I’m not abandoning tech, I’m putting it in service of a different role.
“Am I legitimate?” I’m just starting out, I don’t have a portfolio of shipped AI projects. That’s honest, and I own it. What I have instead: technical understanding that most AI POs lack, and a genuine desire to learn by doing.
What I’m looking for now
Teams with serious AI projects to launch — not POCs to impress the C-suite, but real products that will run in production.
Clients who understand that the value of a good AI PO is avoiding 6 months of dev in the wrong direction.
And projects where my technical background is an asset, not a quirk on a résumé.
If you’re reading this and recognize that need — let’s talk.
Stéphanie Caumont
AI Product Owner · Learn more