Start here — you define product, not code

./product-manager
1

You’re a PM. Your job is deciding what gets built and why. The how is for the developers. Today that division is broken:

  • You write the PRD in Notion. Devs read it once. Three weeks later nobody remembers exactly what it said.
  • You ask for acceptance criteria. You get back “feature works”. When you ask how it was measured, you get silence.
  • Architecture decisions happen in Slack at 11pm. None of them get recorded with their “why”.

The hidden cost: developers spend most of their time on the validation layer — making sure what they built actually does what was asked. That layer is invisible to you. You see the green dashboard, not the work that earned the green.

PaellaDoc inverts this in three moves.

What changes for you

  1. The PRD lives in the orchestrator, not in Notion. Each story has explicit, measurable AC. When you change an AC, the change is timestamped with reason.
  2. Each AC closes with evidence — real Playwright against the dev server with the agent’s code applied. When a dev (or agent) declares a story done, you open the evidence and watch it run. No status meetings.
  3. The work-item view shows what each agent decided, not just what it shipped. When a story closes choosing library X over Y, the why is recorded. Months later when you ask “why did we use X here?”, the answer is in the work-item.

First step

  1. Open PaellaDoc → “new project” → enter as PM (not as dev).
  2. Write the PRD with orchestrator assistance. What matters: define epics with measurable criteria, not epics like “implement dashboard”.
  3. Each story gets assigned to a developer (human or agent). You don’t write code — you review evidence.

Share here

  • What metric you used until now to know if a story was really closed. (Honest answers welcome: “the dev said so” is the most common.)
  • Which architecture decision has bitten you in a past project because it wasn’t recorded at the moment.
  • If you want to validate a PRD you already have, paste it (or its epics) here. Other PMs and devs will comment on which AC are measurable and which are vague.

— @jlcases