Shell
MILO
Status: [ ] Planning | [ ] Active | [x] Shipped | [x] Maintenance Started: 2024-12-28 Shipped: 2024-12-30 (v0.3.0)
One-Liner
MILO is a Signal-to-Noise Life Planner that helps cut through daily noise and focus on goal-aligned actions.
Problem Statement
Eddie experiences constant context-switching between projects, tools, and priorities with no unified view of what actually matters today. Calendars show meetings, todo apps show tasks, but nothing shows the synthesis — what's the one thing I should focus on right now?
Solution
MILO is a desktop productivity dashboard (Pip-Boy themed) that:
- Aggregates signals from multiple sources (activity monitoring, tasks, goals)
- Applies Eddie's "signal over noise" philosophy with GREEN/AMBER/RED state classification
- Provides daily AI-powered Morning Briefings and Evening Reviews
- Detects drift and nudges back to focus
Success Metrics
- Eddie uses MILO daily for at least 2 weeks
- 14-day consecutive use goal (in progress)
- Reduces "what should I work on?" decision fatigue
- PostHog analytics tracking (opt-in)
Shipped Features (v0.3.0)
Core Features (All Complete)
- Goal Hierarchy System (Beacon → Milestone → Objective → Task)
- Morning Briefing (AI-powered daily planning with Claude)
- Activity Monitoring (GREEN/AMBER/RED state detection)
- Drift Detection & Nudges
- Evening Review (Reflection + scoring)
- S/N Scoring System (0-100 with gamification)
- Pip-Boy Desktop UI (CRT effects, glow, scanlines)
- Quick capture with AI parsing
- Stats dashboard with 7-day trend
- Streak tracking with milestones
Not in v1.0 (Scope Fence)
- ❌ Mobile companion app
- ❌ Team features
- ❌ Analog task capture (OCR)
- ❌ Custom AI models
- ❌ Cloud sync (v2 feature)
- ❌ MCP Integrations (Notion, Calendar, Notes) - P1 for future
Technical Decisions
Stack
| Layer | Choice |
|---|---|
| Runtime | Electron 28.x |
| UI | React 18 + TypeScript |
| Bundler | electron-vite |
| Styling | TailwindCSS |
| State | Zustand |
| Database | better-sqlite3 (SQLite) |
| AI | Claude API |
| Analytics | PostHog (opt-in) |
Key Architectural Decisions
- SQLite over cloud DB - Privacy-first, offline capable, simpler architecture
- Electron over Tauri - Better ecosystem, more mature for macOS tray apps
- Zustand over Redux - Simpler API, less boilerplate for this scope
- PostHog for analytics (opt-in) - Privacy-first, EU data storage, self-hostable
Current Phase
Phase: Stage 11 - Listen & Iterate
Goal: Collect feedback from daily use Status: Active since Dec 30, 2024
Tracking
- 14-day consecutive use goal
- First bug reports
- First feature requests
- First analytics insights
Future Roadmap
- P1: MCP Integrations (Notion, Calendar, Notes)
- P2: Mobile companion
- P2: Cloud sync
Files & Links
- Repo: ~/Development/milo
- Release: https://github.com/eddiebe147/milo/releases/tag/v0.3.0
- Pipeline Status: ~/Development/milo/PIPELINE_STATUS.md
- PRD: ~/Development/milo/docs/PRD.md
- Technical Design: ~/Development/milo/docs/TECHNICAL_DESIGN.md
Session Log
2024-12-28 - Project Started
- Stages 1-6 completed in single day
- Full Pip-Boy UI, AI integration, activity monitoring
2024-12-30 - v0.3.0 Shipped
- Stage 7-10 completed
- 355 tests passing
- PostHog analytics integrated
- GitHub Release published
- First user (Eddie) onboarded
2026-01-01 - Pipeline Integration
- Integrated MILO into new pipeline system
- Synced project file with repo's PIPELINE_STATUS.md
Learnings
- Electron + SQLite + Claude API is a solid stack for local-first AI apps
- Pip-Boy theme creates unique identity and dopamine hits
- Morning/Evening dialogues with AI are genuinely useful (not gimmick)
- 6-day build cycle (Dec 28-30) proves focused execution works
Post-Mortem
What Went Well
- Shipped in 3 days (Dec 28-30)
- 355 tests from the start
- Clean Pip-Boy aesthetic
- AI integration genuinely useful
What Didn't Go Well
active-winnative module issues in dev mode (works in prod)- Code signing skipped (need Developer ID cert)
Patterns to Extract
- Electron + SQLite + Claude API architecture pattern
- AI-powered dialogue system pattern
- Pip-Boy theming approach