I’m a PM who builds with AI.
Not just talks about it.
Over 10 years working across London, Singapore, and Vietnam — from compliance and product at PayPal to an independent portfolio of AI products, advisory work, and things built for communities I’m part of.
I supercharge my PM work with tools I’ve built myself: a 16-skill Claude Skills Suite that acts as a thought partner across strategy, prioritisation, roadmapping, and red-teaming; daily use of Claude, Claude Code, and Codex; and MCP integrations that connect AI directly into my workflows. When I say AI-first, I mean it structurally, not aspirationally.
I think in systems, not features. Before any build decision, I’m asking: what does this connect to, what does it break, and what would I have to believe for this to be wrong? I consider trade-offs explicitly — speed vs. quality, breadth vs. depth, building vs. learning — and I document them so future me (and collaborators) know why we made the call we did. I build evals into AI products from the start, not as an afterthought, because a model you can’t measure is a model you can’t improve. And I ship scrappy: the smallest useful thing, then watch what real users do with it. I encourage everyone I work with to do the same — waiting for perfect is the most expensive decision in product.
I’m a practitioner of Insanely Simple — the design philosophy Ken Segall documented from his years working with Steve Jobs and Apple. Simplicity isn’t a style choice. It’s a discipline: reduce to the essential, then ask if you can reduce further. It shapes how I write PRDs, how I evaluate user flows, and how I push back on feature creep.
Product sense is the thing AI can’t replace — knowing which problem is actually worth solving, reading what a user means versus what they say, and making the call when the data is ambiguous. I build closest to problems I’ve witnessed firsthand, which is why Red Dot Times, Mochi’s World, Little Feelings, and Mosaic Mandarin all came from real friction I observed in communities I’m part of. The most unforgiving user feedback I’ve ever received came from kids. You can’t ship mediocre to a five-year-old.