Turn your neighborhood into a daily adventures map with a cozy AI companion 🌸
Description and use
Outside.exe shows how a local Gemma 2B model can replace conventional habit trackers by offering real‑world quests and rewards for offline activity.
Problem
Users of productivity and habit‑tracking apps feel overwhelmed by dashboards and push notifications, leading to doomscrolling instead of real‑world breaks.
Outcome
Outside.exe demonstrates how a lightweight local LLM can replace push‑driven habit trackers with gentle, real‑world nudges, achieving offline engagement and user‑controlled data privacy.
Workflow
- 1. User opens app and sees a frosted‑glass check‑in modal.
- 2. User inputs mood, energy level, and available time.
- 3. Frontend sends data to FastAPI backend.
- 4. Backend constructs a strict prompt and calls local Gemma 2B via Ollama.
- 5. Gemma returns JSON with quest title, description, steps, and XP reward.
- 6. Frontend parses JSON and displays quest board with tasks.
- 7. User completes tasks, optionally taking inline camera photos per step.
- 8. XP is awarded, companion animates, and progress is stored in LocalStorage.
Evidence
Outside.exe flips the game loop upside down: it uses open‑weight AI to turn the physical world into an offline questing ground.
The backend calls Google's Gemma 2B (gemma:2b) running locally via Ollama.
I learned about event bubbling and used e.stopPropagation() on the camera button to separate the actions.
- Record date: 2026-10-10T21:10:58.669Z