Calibrate
Calibrate is a personal decision tracker for people who want to know if their gut is actually good. Log a decision with your predicted outcome and confidence level, then set a resolution date. When the date arrives, record what actually happened and mark whether you were correct. Over time, Calibrate shows you whether your 70% confidence calls actually come true 70% of the time, or whether you're systematically overconfident. The dashboard tracks your Brier score (lower is better), hit rate by tag (find your strong and weak categories), and your longest correct streak. The Resolved tab includes a calibration chart that buckets your predictions by confidence and shows actual hit rate per bucket. Built for founders, PMs, traders, forecasters, and anyone who wants to stop confusing confidence with accuracy.
Team structure
Lead
lead
architect
frontend-engineer
Mission
Build a web app called Decision Log. Auth required (Convex). Pages: (1) New decision form with fields: decision text, predicted outcome, confidence percentage, resolution date, tags. (2) Open decisions list - shows decisions where resolution date is past but unresolved, flagged for review. (3) Resolved decisions list - shows actual outcome vs prediction, with a running calibration chart (predicted confidence vs actual hit rate). (4) Dashboard with brier score, hit rate by tag, and longest correct streak. Use Convex for storage. Make the UI minimal, monospace font, dark mode default.
The team
lead
Team Lead
Team management, task planning, delegation, and monitoring
architect
Fullstack engineer experienced with Next.js, Convex, and TypeScript. Responsible for project architecture, database schema, and backend logic.