Compass Decisions Team designs and builds AI-powered planning tools that remove the friction between deciding what to do and actually doing it. Our flagship product is a mobile-first, voice-driven web application that listens to a task, breaks it into small, concrete, time-estimated steps, and schedules it — with minimal effort from the user.
This team hosts the development of that product, built in partnership with our client, Roberto.
The core loop is simple by design:
- Speak — a voice prompt, a brain-dump of the day, or a vague goal ("finish the report by Friday and clean the garage").
- AI breaks it down — complex or messy input is parsed into small, concrete, time-estimated chunks.
- Scheduled automatically — chunks are written to the user's Google Calendar after a lightweight, one-tap confirmation. The calendar stays the system of record; we don't replace it, we organize around it.
Planned next: Google Tasks and Google Classroom as additional input/output sources, so the same chunking engine can pull from and write to more of where users already plan.
| 🎙️ Voice-first, touch-fallback | Every core action is completable by voice; every voice action has a visible touch equivalent. |
| 🧠 AI task chunking | Turns unstructured, spoken goals into small, schedulable steps — not a rigid command syntax. |
| 📅 Google Calendar sync | Reads existing calendar context and writes new events, after confirmation. |
| 🔜 Google Tasks & Classroom | Planned integrations to widen input sources and sync targets. |
| 📲 Progressive Web App | Installable, mobile-first, with push notifications for reminders and schedule nudges. |
| ⚡ Low-friction confirmation | One tap or word before anything is written to the user's calendar — never a form. |
Living reference — the authoritative, up-to-date stack lives in each repository's own README. This is the current high-level direction.
| Layer | Approach |
|---|---|
| Frontend | Mobile-first Progressive Web App |
| AI / Orchestration | LLM-based task parsing and chunking |
| Voice | Speech-to-text input, text-to-speech feedback |
| Calendar & Tasks | Google Calendar API (live), Google Tasks & Google Classroom APIs (planned) |
| Notifications | Web Push via the PWA |
- Phase 1 — Product definition & UX/UI design
- Phase 2 — Frontend build (responsive PWA)
- Phase 3 — Google OAuth + Calendar integration, backend & data model
- Phase 4 — AI pipeline (speech → structured chunks → schedule proposal)
- Phase 5 — Production hardening: testing, monitoring, accessibility audit, beta
| Repository | Description |
|---|---|
.github | This repository — organization profile and org-wide community health defaults |
app | Main application (rename to match your actual repo) |
docs | Product, architecture, and API documentation |
Keep this table current — it's the fastest way for a new teammate or reviewer to orient.
This is a private client engagement. Team members should follow the branching strategy, commit conventions, and PR review process in CONTRIBUTING.md. Access to source repositories is by invitation only.
Please do not open a public issue for security concerns. See SECURITY.md for our responsible disclosure process and contact.
Maintained by the Compass Decisions Team. For project inquiries, open a discussion in this organization or reach out to the maintainers directly.