Open your notes and write two columns: Chat/prompt volume metric | Checkable work metric. Fill in 2 rows for your own job. You're done when you've named the one outcome metric you'll track this week for an AI tool, in 12 words or less.
How-to · ORG · ORG.3
AI tool value = checkable work shipped, not chat volume
When a team claims an AI tool 'works,' ask what checkable work shipped (PRs merged, tickets closed, docs published) — not how many chats or prompts were sent.
Worked example
Simon Willison notes that Claude-via-Slack style workflows can genuinely land real pull requests. The right unit of progress is merged, reviewed work — not how many prompts you sent. The builder-track skill is using AI tools to actually ship, then measuring the outcomes. You still own merge and send.