Same model, five angles
Same idea as the one-pager, cut five different ways. Pick the angle that matches your question and go straight there.
The repo holds your org's steering documents, and the AI coding agent reads them before it generates anything — so every prototype, spec, and requirement it produces already follows your conventions. Full Overview →
The old model has the PM speculate in a document, hand it off, and wait to find out if the speculation was right. ProductHarness lets the PM build a working prototype and validate it with real stakeholders in the same session they're thinking through the problem — so what ships is derived from evidence, not a guess, and engineering inherits something built to standard instead of a document to interpret. Framework →
The PM supplies intent, judgment, and the definition of “good.” The agent supplies the mechanical translation — code, requirements, acceptance criteria, test cases — all shaped by the org's standards. Nobody's accountability moves. Roles →
Not a requirements doc, not a separate PM tool — the same developer environment engineers already work in, with the agent present the whole time. That's not a preference; it's what makes the model work. Setup →
One project repo gets created per initiative, the agent sets everything up from there, and what the org can do with it grows — from day-one specs and prototypes toward outcome-tied requirements and beyond. A path, not a switch. The Adoption Path →