Writing
Writing on AI product strategy, evaluation, adoption, and what is coming next.
Where AI belongs in a product, and the difference between a feature that compounds and one that competitors copy in a quarter.
AI alone will not differentiate a product, and AI tool usage is not a strategy. The three mistakes that most often put a product team behind: copying competitors, chasing low hanging fruit, and overestimating how much inaccuracy customers will forgive.
Read the pieceWhat has to be true before an AI product deserves a customer's trust, and how teams prove it rather than assert it.
Without evals, quality is an opinion and trust is a guess. What two bad months cost Claude Code, and the five ways weak evaluation shows up on the business side of an AI product.
Read the pieceProduct craft, human judgment, and the conventional wisdom about building with AI that is worth arguing with.
Two studies show the same generative AI tools that lift individual output also homogenize the ideas a team produces. Mandating usage without room for judgment quietly erodes the differentiation a strategy depends on.
Read the pieceWeak signals, emerging capability gaps, and how to act on a long-range view while it still counts as an advantage.
Some companies are already outpacing competitors by building agents into real workflows. The gap between adopters and the rest is widening quietly, and late movers feel it compounding.
Read the piece