Stores decisions with explicit assumptions and review triggers, then classifies them as valid, at risk, or invalidated when new evidence arrives. I used Codex with GPT-5.6 to implement the ledger, trigger evaluator, tests, and product experience. A future live path can use GPT-5.6 to extract candidate assumptions; the public demo is explicitly a precomputed, tested fixture, and owners approve the final assumptions and triggers.

Built With

Share this project:

Updates