Everything, organized by how you read.
The Narrative You Don't Have to Teach
A borrowed clinical narrative did more for our product story in one week than a quarter of feature announcements. On finding the story your users already believe.
They Can't Ask for What They Can't See
Trial feedback keeps finding users who never knew the feature existed. The fix we're testing: 30 to 90 narrated seconds of the product doing real work.
When the Assistant Becomes the Navigation
A designer asked the fair question about a bet that changes what a surface is: where's the revenue, and what comes after? A case for grading paradigm bets differently.
Lead with the Risk
Crit on a two-factor enrollment flow found 3 copy sins: a euphemistic link, a warning that opens with praise, and a screen that never names what it enrolls you in.
Start with the Smallest Room
We changed who's in the room for early AI discovery: the experience owner and the model expert, nobody else at first. A hackathon proved the pattern.
Philosophical Daylight
A hiring panel reached yes and then spent the hour naming the tension that came with it. Why the named flag is the most useful artifact a debrief produces.
Say No with a Name Attached
Coaching a lead whose superpower is saying yes to everything. The fix is a redirect: not just no, but here's who can run with it.
The Answer Got Designed in the Room
A crit on a small client-facing setting ended with a live redesign built from a pattern the product already owned. How the room got there, and what it says about running crit well.
Ship It in One Room First
We decided to ship a new navigation pattern in one part of the product over honest dissent about long-term inconsistency. The reasoning, the data, and how the dissent shaped the rollout.
Prioritization Isn't Real Until the Swap Is Named
A harsh comment in an executive review turned into a week of thinking about trade-offs, who names them, and why addition without subtraction reads as capacity.
20 Minutes and $5 per Frame
One of our design leaders handed an AI a Figma frame and asked for design critique. The results draw a precise map of where AI design review works today, and where it doesn't yet.
From Review It to Break It
A one-word change to how you run crit: ask the room to break a working prototype instead of reviewing flat screens. The mechanics, the precedent, and a starting point for your team.
AI Development Starts with Discovery
AI inverted the economics of product development: demos got cheap, verification stayed expensive. A case for moving discovery to the front and putting an evaluation gate before build.
Protect the Phases
Early feedback that means to save time is usually the feedback that costs the most of it. On why phases deserve a doorman, and why the leader has to make the pushback cheap to say.
Turn-Down Service
The whole industry answers every AI product question with a chat window. For clinicians who spend their day in conversation, that's the wrong gift, and a case for ambient help over one more thing to talk to.
Skip the Courier
Two leaders asked me to carry their reorganization cases up the chain. Why I declined both, and why the refusal did more coaching than any case I could have delivered for them.
The Cheap Conversation
Strip the tooling away and my best AI habit is just briefing, the oldest management technology there is: spend the hour where mistakes are free, then delegate.
Two Spikes, One Role
We ran two finalists through an identical interview spine and still nearly picked wrong, because one panelist was measuring against a different job. What the debrief is really for.