Product Manager Hub
Join the free weekly email, or share a decision or an insight.
Illustrative example, not a real submission
Situation: A small team could build its own onboarding flow or buy a tool.
Options I considered: Build, buy, or delay a quarter.
What I decided: Delay and test the current flow first.
What happened: The current flow was enough for the people already using it.
What I would do differently: I would watch people use the current flow before I buy or build.
One short email a week, after you confirm.
A library of real decisions and insights I have approved.
MCP lets you search the library from inside your own Claude.
Share a decision or an insight. I review it before it is published.
I collect real decisions and insights from product managers and product owners.
I publish the useful ones.
I send one short email a week.
One short note. The same sections I use when I write the week.
Format of the weekly email, not one I have sent
A decision is the situation and what happened. An insight is one thing learned, and what it is based on.
Illustrative example, not a real submission
Situation: A team was deciding whether to let an AI assistant draft first versions of specs.
Options I considered: Let the assistant draft the spec, keep writing specs by hand, or use the assistant only after a draft exists.
What I decided: Let the assistant draft the first version, then a person rewrites it.
What happened: The draft was faster to start and still needed a person to make it specific.
What I would do differently: I would not ship the first draft the assistant wrote.
Illustrative example, not a real submission
What I learned or noticed: Asking an AI tool to critique a draft works better than asking it to write the draft.
I tried it: I tried it on my own drafts.
Illustrative example, not a real submission
What I learned or noticed: A roadmap review goes better when each item says what would change the decision.
I saw it happen: I saw it on my team.
Why confidence, not capability, is what deep AI products compete on. A first-person look at two-sided drift, rented models, and the measurement you actually own.
Read article
AI-assisted code works in demos. It stalls in production. Here's why the gap exists and how to close it.
Read article
MCP isn't a technical detail—it's a strategic decision about whether your AI product can evolve independently of your model vendor. Here's how to think about it.
Read articleFAQ
Straight answers about the weekly email and the library.
