~/jaredclark · /pm-os
accent
display
● live

Product / how I work

AI didn't change
the work.
It changed the leverage.

// the role

Most PM work is the same questions, asked again next quarter, with last quarter's answers lost in someone's notes. The system below is what I built to stop the reset.

// the discipline

That means being close enough to the customer to make real tradeoffs, and disciplined enough to build systems that compound learning over time, instead of starting from scratch every quarter.

// the leverage

AI is infrastructure, not a shortcut. Used well, it dissolves the process overhead that eats a PM's week, freeing time for the work that actually requires a human: judgment.

01 the system

A workflow that compounds, not one that resets every Monday.

~/pm-os / claude-code running
Built on Aakash Gupta's PM Operating System as a foundation.

An AI-powered PM workflow that runs a Net Product Value loop. It's Net Present Value, applied to product instead of capital: identify → define → build → measure → iterate → actualize. The discipline: ask "is this still worth it?" at every stage.

Running in Claude Code: a persistent context library (company strategy, stakeholder profiles, PRDs, decisions, live metrics); skills exposed as slash commands that auto-route to relevant context; sub-agents that critique work in parallel from different angles such as engineering, UX, executive, customer, stakeholders; all before anything ships.

Claude Code
orchestrator · reads all layers
context loaded · 3.2k tokens
Context persistent
  • company strategy
  • stakeholder profiles
  • active PRDs & decisions
  • meeting + decision log
  • writing style guide
  • estimate-vs-actual log
  • live metrics (MCP)
Skills /slash commands
  • /daily-plan
  • /meeting-notes
  • /prd-draft
  • /impact-sizing
  • /prd-review-panel
  • /status-update
  • /slack-message

+ dozens more

Sub-agents 5 agents · 7 reviewers
  • full-stack engineer
  • ux designer
  • executives
  • legal
  • UX researcher
  • stakeholders
  • customer voice

runs on command

02 a day in practice

Real decisions, under pressure. not a product demo.

9:00am
/daily-plan
Start of day
beat 1 / 7 sets context for the rest of the day
auto-advance · paused

03 what i bring

The system is replicable. The judgment is not.

This is what daily PM work looks like when you stop treating AI as a chatbot and start treating it as a system you design and orchestrate.

The system is replicable: context library, skills, sub-agents, the loop. The rest of the formula isn't.

Judgment × Effort × Ownership = Value

The × matters. Zero out any multiplier and the whole thing collapses. Judgment without effort ships nothing. Effort without judgment ships the wrong thing. Both without ownership stall in committee.

Judgment to know which context to maintain and when to override the model. Effort to run the loop deliberately, every day. Ownership to see each decision through to real value. That's what I'd bring on day one.

Jared Clark / product