Short, practical pieces on building products, putting AI to work, and leading teams that ship. 21 posts since 2023.
Public leaderboards rank models on tasks that look nothing like yours. Here's how I actually evaluate Claude, OpenAI, and Gemini for production use.
Read the postProduct management teaches you how to make a decision well. Founding teaches you what it feels like when there's nobody above you to catch the ones you get wrong.
Read the postBenchmarking, routing and caching across four AI products. What actually moved the bill, and what didn't.
Read the postSwap the model all you want, if retrieval hands it the wrong paragraph confidently, no amount of model quality fixes what happens next.
Read the postLessons from moving a lifestyle app to freemium and subscriptions, and why onboarding mattered more than price.
Read the postSelf-serve won every growth argument I made for years, until I built a product where removing the human was the thing actively costing us trust.
Read the postWhat launching a car marketplace in the UAE and Saudi Arabia taught me about local buying behaviour.
Read the postThe real risk in a legacy migration was never the new platform failing. It was the four months in between, where both systems had to be true at once.
Read the postScaling from a handful of engineers to a 17-person AI product team meant unlearning the habits that made me good at shipping, personally.
Read the postClintwin/Medsim AI isn't a chatbot with a medical skin on it. The entire design brief was about what happens when a model is convincing but wrong.
Read the postMost teams running 'agile' are running the ceremony with the actual agility quietly removed. Here's how to tell, and what to rebuild instead.
Read the postFour AI products, four different markets, one portfolio strategy. The hard part isn't picking a metric per product, it's resisting the urge to.
Read the postThe decision to go freemium took an afternoon. Making it work took a rebuilt product, forty interviews, and a year.
Read the postThe best PM hire I ever made had never held the title. The signal I was actually looking for doesn't show up on a resume at all.
Read the postScoring frameworks don't resolve prioritization debates. They just launder someone's opinion through a formula that looks objective.
Read the postSurveys scale. Interviews don't. For the decisions that actually matter, that trade-off runs the wrong way more often than product teams admit.
Read the postPLG became a label teams reach for before they've done the one thing it actually requires: letting the product sell itself.
Read the postThe hard part of a 90-day MVP was never the 90 days. It was agreeing, out loud, on what we were willing to be embarrassed about at launch.
Read the postThe five stages of design thinking were never meant to run once, start to finish. Treating them that way is where most teams get stuck.
Read the postNobody hands you a product manager title. You just start doing the job, badly at first, until someone notices you already are one.
Read the postThe interview-prep framework everyone memorizes and almost nobody uses on the job, until you actually try it.
Read the post