What is a deterministic seed?
A seed is the short string a generated company starts from. Regenerating from it gets you close to the original. The reproducible artifact is the saved company, pinned by its ID.
Why it matters
Reproducibility is the difference between a test and a guess. If the data changes between runs, a failure tells you nothing. A seed gets you a close regeneration; a saved company, pinned by ID, gets you exactly the same data every time, so the only variable in CI is your code.
How it's used
Generate a company and save it. It gets a short ID like fix-7kq, and Fictix stores the data. Anyone — a teammate, a CI job, a sales engineer next quarter — reads that same company by ID. Use the seed when you want a fresh company close to the original; use the ID when you need exactly the same data. That makes regression deltas trustworthy: same inputs, so a score change is a real change.
Questions
Is a seed the same as a snapshot?
No, and the snapshot is the one to pin. A seed is the input; regenerating from it gets close, but not identical. A saved company (stored by Fictix, with a short ID) is the reproducible artifact everyone replays exactly.
Does the same seed work across environments?
For exact data on your laptop, in CI and in a demo, pin the saved company by ID; Fictix stores it, so every environment reads the same data. Regenerating from the seed gets close.