Writing

Notes on building AI products.

Opinions formed by shipping, not by reading about shipping. Mostly about the gap between what AI can demo and what it can sustain.

Subscribe via RSS

  1. Discovery is the real AI bottleneck — not engineering

    Teams obsess over shipping AI faster while the 60-day discovery cycle in front of it goes unexamined. Here's how we cut ours to 40 days, and what broke along the way.

    • AI Product Management
    • Product Discovery
    • Operating Model
  2. Your AI roadmap is a hype document until it survives unit economics

    Most AI feature lists are prioritised by novelty. Here's the filter I run a $100M+ portfolio through instead — and the four questions that kill most AI ideas before they reach a sprint.

    • AI Product Management
    • Product Strategy
    • Unit Economics
  3. Vibe coding isn't a shortcut. It's a new form of product evidence.

    Product managers who prototype with AI tools stop arguing from opinion and start arguing from artefacts. That changes who wins the roadmap conversation.

    • AI Product Management
    • Prototyping
    • Vibe Coding