Case studies from real work. Insights on ideas worth working through.
A practical guide for product leaders who want real AI workflows now, without becoming an AI expert first.
Real products. Real workflows. Real impact.
Why a student who changes schools can become a stranger to your data, and how the same fix made a first AI feature possible without exposing any PII.
How an AI-built clickable prototype took a homepage from a three-day design estimate to two hours, surfaced a layout problem no spec would have caught, and cut a month of design and feedback down to a week on a dashboard.
Becoming the source of truth when the platform couldn't be.
How I defended roadmap capacity for stability and scale work while the business grew 7x, and why the unglamorous epics are the ones the growth stood on.
How I built a custom GPT to automate release notes, trained it on ten golden examples, then retired it for a native Rovo workflow inside Jira.
Ideas, lessons, and perspectives on AI and product leadership.
Before promising an AI outcome, validate that the data can support it. What querying a platform I couldn't trust taught me about data readiness.
Oct 2026A PRD success metric can support an OKR without copying it. Line of sight is the hypothesis connecting the two, and most PRDs never write it down.
Sep 2026An unvalidated bet is not automatically a bad bet. It is a bet with a wide range of outcomes, and it should be sized like one.
Sep 2026A long-range plan is not a prediction. It is a decision about which future gets your attention today.
Aug 2026AI output is only as good as the input it works from. The PM is usually the closest available source of ground truth in the room, and that turns out to matter more than being a reviewer.
Aug 2026How a spoken bug report becomes a paste-ready defect ticket, using the same AI-interrogates-you method behind the Prompt Blueprint, applied to specs instead of mocks.
Aug 2026Why talking through a spec out loud, before typing a word, produces better product thinking.
Jul 2026Context, Role, Discovery, Execution: the four-part framework I use on every AI task, illustrated with the real questions I run before building a single product mock.
Jul 2026No articles match those filters.