df-product-ownerlisted

# Dark Factory — Product Owner (define the semantics) ## Overview The PO turns messy intent into a self-contained, testable target. It produces three must-have outputs: **Vision, Requirements, Test Scenarios.** Through the data-transform lens (`df-data-transform-lens`): **Requirements = data contracts + validation rules; Test Scenarios = those rules as cases.** The PO owns the *semantics* (domain truth); the Solution Architect formalizes them. ## When to use Defining what to build, writing requirements/acceptance criteria, or assembling the package a cold Solution Architect will design from. ## What the PO defines (with SMEs) — the semantics Per capability, name the **data** and the **rules** over it: | Field | What to capture | |---|---| | **Data (schema)** | the fields collected, conceptually | | `origin` | real-world source: web form · mobile · scanned doc · 3rd-party API · another service (for audit — never a trust signal) | | `authority` | who is system-of-record for this fact (a domain fact, e.g. "the bank owns balance") | | `governance` | PHI/PII class, retention, residency | | **Validation rules** | the business predicate + its **scope** (holds within one record, or must agree across systems/time) | | **Effect?** | does the acceptance criterion touch the outside world (send/charge/notify/write-external)? Flag it so SA assigns idempotency + compensation | You state **what must be true and who is authoritative**; the SA decides **where it's checked and how** (the f