gale
A local-first personal operating system and CLI agent built for JEE preparation. Replaces rigid task apps with conversational logging, persistent memory, and capacity-based planning.
Context
Preparing for JEE (Main & Advanced) over two years is essentially a systems problem: 90+ chapters across Physics, Chemistry, and Math, non-stop coaching schedules, test series, and constant backlog risk.
Standard tools fall apart quickly. ChatGPT has amnesia and invents new plans every morning. Notion feels like a spreadsheet job after a 6-hour coaching day. Generic planners create impossible 14-hour schedules that collapse the moment you miss a day.
I built Gale (named after Gale Boetticher from Breaking Bad, for the obsessive precision and clean lab notes) to solve this.
How It Works
The foundational rule of Gale is simple: the AI is not the memory; the AI operates the memory.
All chapter metrics, question completion counts, accuracy rates, and prerequisite chains live locally in structured files on disk. An autonomous CLI agent reads the state, calculates constraints, and updates records deterministically.
- Conversational logging: Instead of editing tables, I log updates naturally (“did 32 NLM questions from Step 1, 26 right, struggled with pulley constraints”). Gale parses the text, updates subject metrics, and logs concept traps in a single turn.
- Capacity-based planning: Plans are budgeted using real solving speeds and actual daily free hours, differentiating school days, coaching days, and test weeks.
- Drift recovery: If a day is missed due to illness or fatigue, Gale absorbs the drift instead of dumping work onto the next day and causing an avalanche.
Status
Currently in private beta and active development.