← projects

may.os

A personal operating system through WhatsApp

Most personal tracking tools die the same way - you download an app, fill forms for a week, and quietly stop. The friction kills the habit.

This project starts from a different premise: what if the interface was a conversation you're already having? WhatsApp is where I think out loud, share meals, vent about work, celebrate wins. Instead of asking me to open another app, the system meets me where I already am.

A single message - text, voice, or photo - gets classified by AI into one of six life areas. No forms, no categories to pick, no buttons to tap. Just talk, and the system understands.

How it works

input
WhatsApp text, voice, photo
process
AI classifies Claude Haiku / Sonnet
store
6 life areas Supabase
Dashboard Weekly Pulse Journey

What it captures

Nutrition

active

Meals, macros (kcal, protein, carbs, fat), daily progress, nutritional tips. Photo analysis for food identification.

Body

planned

Workouts, sleep quality, weight trends, physical energy. Movement frequency over intensity.

Emotional diary

active

Energy level (-1 to 1), emotional tone (16 categories), recurring themes, celebrations. One entry per day, appends naturally.

Work

planned

Accomplishments, blockers, deep work sessions, skills developed. Wins over hours.

Relationships

planned

Meaningful connections, quality of interactions, relational patterns. Only when it matters.

Intellectual

planned

Readings, ideas, insights, frameworks learned. Feeds from the vault.

The rhythm

Daily - capture

3-5 natural WhatsApp messages throughout the day. No forms, no checklists. AI extracts the data. The bot occasionally asks - "how did you sleep?", "what did you accomplish today?" - but never more than twice a day.

Weekly - pulse

Every Sunday, the bot sends an AI-generated summary: nutritional adherence, movement days, average energy, dominant tone, recurring themes, wins, and one cross-area correlation. Not a list of numbers - a mini-narrative.

Monthly - deep dive

Energy curves, mood heat maps, consistency radar across all six areas, month-over-month comparisons. The editorial graphs that tell the story of a month lived.

Under the hood

TypeScript Astro Supabase Claude API Twilio Vercel Groq Whisper

What's next

  • Dashboard with editorial graphs
  • Body and movement tracking
  • Proactive check-ins from the bot
  • Weekly Pulse summaries via WhatsApp
  • Work and intellectual area capture
  • Monthly deep dive with cross-area correlations

In progress · Started March 2026