Path 02
You have skills that transfer — engineering, domain knowledge, consulting, data. What you need is a clear line from where you are to your first AI PM role.
The path into AI product management is not a degree. It's not a bootcamp certificate. It's a credibility argument — and you build that by demonstrating that you can think clearly about AI systems, understand the tradeoffs, and translate between the people who build them and the people who use them.
The transition is faster than you think if you work on the right things. Most aspiring AI PMs spend their time learning the wrong things — ML theory, broad surveys, Twitter threads. What actually matters is frameworks for decision-making, a working vocabulary for technical conversations, and at least one domain where you have real depth.
Everything below is ordered by what I'd tell someone starting today. Skip what you already know.
Where to start
Five steps, in order. Not a curriculum — a credibility-building sequence.
You don't need to code models, but you need to talk credibly about how they work. RAG vs fine-tuning, inference costs, hallucination failure modes, evaluation metrics. This is the foundation that makes every other conversation possible.
Generalist AI PMs are commodity. Domain-specific AI PMs — healthcare, fintech, legal, manufacturing — are rare and valuable. If you already have domain experience, that's your moat. Leverage it. If you don't, pick one and immerse yourself for 6 months.
PRD writing for AI products is different. Prioritization in probabilistic systems is different. Metrics for AI features require new thinking. You need frameworks that acknowledge the nondeterministic nature of what you're building — not frameworks copied from SaaS product management.
Side project, open-source contribution, internal AI tool at your current company. It doesn't need to be impressive. It needs to exist. You're demonstrating that you can take a fuzzy problem, make decisions under uncertainty, and ship something.
The AI PM interview is a product sense interview with an AI overlay. Study the 50 questions that actually get asked. Practice structuring answers around the AI-specific tradeoffs — data quality, model evaluation, trust and explainability — not just standard PM frameworks.
Required reading
Highest-signal articles for the transition — ordered by where most people get stuck.
Structured learning
Practical courses built from real experience — not recycled textbook content.
If you're mid-transition and have a specific question about positioning, skill gaps, or what to build — send me a note. I read everything, even if I can't respond to everything.