ModelLedger

ModelLedger is a decision journal built specifically around mental models. You log a decision in 60 seconds: context, model used (e.g., inversion, opportunity cost, second-order effects), confidence, and what you expect to happen. The app then schedules lightweight follow-ups (1 week, 1 month, 3 months) to capture outcomes and compare them to your predictions. Over time it generates a “model accuracy” dashboard showing which models correlate with better outcomes for you, in which domains (work, money, relationships), and where your biases show up (overconfidence, recency, sunk cost). It’s not therapy and it won’t magically make you smarter; it’s a structured feedback loop most people never build. The product is intentionally boring: fast logging, reminders, and clear analytics. If users don’t review outcomes, it provides no value—so the UX is designed to make review unavoidable and quick.