🎤 Talks
Things I've stood up and said out loud.
Three talks so far. The web decks keep their figures interactive; the Version One workshop also includes a downloadable PDF from the live event.
-
Every Time It Was Wrong
Introducing the PR reviewer to the team. What it does to your pull request, why it says so little, what happens when you reply to it, and the measurement that surprised us: its comments were right almost every time, and where they were printed decided whether anyone acted on them. See how each comment was scored.
-
Everybody Wants the Agent. Somebody Has to Build the Floor.
How an organization goes from developers holding AI subscriptions to real automation, in five stages over eighteen months. Skills before automation, and the coder built last, once the four skills that supervise work can catch it being wrong. The evidence is the public research; the failure modes are mine.
-
Ten Times Zero Is Still Zero
A talk for students about working with AI. It multiplies what you already are, so become worth multiplying first. The receipts come out of git, the playbook is stealable, and the figures are live: drag the multiplier to zero, run six sessions with and without the loop and watch the bill arrive, or open a cold session with no brief and see what a model does first.
Prefer to read them? Two of the three have a post behind them, with the same argument and the receipts written out: Everybody Wants the Agent and Ten Times Zero Is Still Zero. The written version of Every Time It Was Wrong is still being drafted.