The system behind it
How I keep it all moving.
The tools and operating habits I use to run several products in parallel without losing track. The two apps below run on a scrubbed snapshot of my real slate and roadmap; anything you change stays in your browser.
Build Slate
One board for every product: stage (idea, spec, building, beta, launched, paused), the very next step, and a milestone date. It caps active projects at five, which forces depth over breadth.
Near-term roadmap
Later, Next, Now and Done lanes with a board per product, owners, work-session estimates and a blocked flag. Any AI session can read and update it by link, so it stays current without manual upkeep.
Source for both apps, plus the PM templates and routines below, is on GitHub.
Decisions that leave a trail
- Decision log: every call records why, and what it beat.
- Persona check: before a product decision is final, each persona gets a green or red with a one-line reason. CI fails if a decision file skips it.
- Anchors: a few principles every feature is checked against, like "anxiety-less" and "stays on US soil."
- Post mortems: each one adds a rule, so the same mistake doesn't happen twice.
An AI team with lanes
Specialist agents each own one job: staff PM, QA (tests first), engineer, privacy and security, UX for low-tech users, and marketing. I keep an inventory of every agent's maturity (proven, working, dormant or broken) and decide which to use, mature or retire.
One hub across projects
Each project reports into a central product office: products, PM tools, agents, decisions and open loops, so nothing lives only in one chat. The story of how that came together.