Pier Madsen

Tenex sells the roadmap I run in-house. Where we match, and the step I'd add after the final presentation.

Results from the first find, prove, scale project nine months in, and why the monitoring layer turned out to be the bigger asset.

None of the teams I work with report to me. Notes on getting workflows changed anyway.

The clever retrieval experiment gave way to plain markdown files, and it works better.

A new job where AI is the whole job, and the first thing I've had to unlearn.

Karpathy named the way I already build my spare-time tools. Here's where I do it and where I don't.

The industry is standardizing the thing I've been hand-building for two years.

After a year of hand-feeding custom GPTs, I finally tried retrieval. Notes from the first rough version.

Managing six people and maintaining custom GPTs turned out to be the same lesson about context.

Mollick's four rules for working with AI, checked against a year of doing it daily.

Ng's four design patterns describe AI a step past where I am, and one of them keeps bugging me.

What I expect from my team, and from myself, now that a lot of our first drafts come from AI.

My reaction to Lenny Rachitsky's take that AI will come for the high-level PM skills before the low-level ones.

Cagan's new book on the product operating model, read from inside a company that isn't a tech company.

What I've learned about winning resources for a product when every decision-maker is measured on something different.

A three-month reflection on launching a recruiting platform across 26 offices.

Why a Lenny’s Podcast conversation about product management resonated with me in 2023.

The questions I use to clarify a product problem, test assumptions, and prepare a PRD before writing requirements.