Skip to content

Roadmap

Jump to bottom
WilliM233 edited this page Mar 29, 2026 · 1 revision

Roadmap

BRAIN 3.0 is built in phases. Each phase adds a capability layer to the system without requiring changes to previous phases.

Phase 1 — The Core Loop ✓

Status: Complete (v1.0.0, March 2026)

The data foundation and conversational interface. Claude has full read/write access to all seven pillar entities through the MCP.

What's Delivered

  • PostgreSQL database on Docker with all seven pillar tables and relationships
  • FastAPI REST API with full CRUD, composable query filters, and four reporting endpoints
  • Claude MCP integration via brain3-mcp — 50+ tools covering all entities
  • ADHD-aware task metadata — energy cost, cognitive type, activation friction, context requirements
  • Routine system with streak tracking, flexible scheduling, and pattern break detection
  • Activity logging with subjective metadata (energy before/after, actual friction, mood)
  • Reporting — activity summaries, domain balance, routine adherence, friction analysis
  • Production deployment on TrueNAS with Docker Compose, automated migrations, nightly backups
  • CI pipeline — 283+ tests, Ruff lint, GitHub Actions
  • Documentation — README, developer setup, deployment guide, environment reference, wiki

What's Not Here

  • No web UI — interaction is through Claude conversation only
  • No authentication — single-user system behind a firewall
  • No scheduler — the system responds but doesn't initiate
  • No push notifications or voice interface
  • No Home Assistant integration

Phase 2 — The System Gets Proactive

Status: Planned

The system transitions from responsive to proactive. Claude reaches out rather than waiting to be asked.

Planned Capabilities

  • Internal scheduler (APScheduler or cron-based) for time-triggered behaviors
  • Home Assistant integration — push notifications via the companion app, voice interface via Assist
  • Morning briefings — daily summary of what's on deck, routine status, and relevant context
  • Accountability nudges — missed routine alerts, deadline warnings, neglected domain flags
  • MCP composite tools — higher-level operations that combine multiple API calls (e.g., "complete routine + log activity + check in" as a single tool)
  • Modular MCP refactoring — splitting the monolithic server.py into per-entity tool modules as composite tools justify the structure

What This Changes

Phase 1 BRAIN waits to be asked. Phase 2 BRAIN notices things and reaches out — a missed routine, an approaching deadline, an area of life that hasn't gotten attention in weeks. The ADHD brain struggles with object permanence; Phase 2 addresses this by making the system visible through the channels you already check.

Phase 3 — Browse and Manage Directly

Status: Planned

A lightweight web UI for when you want to see your world at a glance without having a conversation.

Planned Capabilities

  • Mobile-first web dashboard — goals, projects, tasks, routines at a glance
  • Quick entry — add tasks and log activities without a full conversation
  • Visual reporting — charts and trend lines for pattern recognition (energy cycles, routine adherence, friction analysis)
  • Authentication — API key, OAuth, or basic auth middleware (required once the UI exposes the system beyond localhost)

What This Changes

Phase 3 adds a visual layer that complements the conversational interface. Claude remains the richer interface for reasoning and pattern interpretation; the UI provides at-a-glance awareness and low-friction data entry.

Phase 4+ — Expand and Integrate

Status: Vision

Everything beyond Phase 3 is speculative. These are directions, not commitments.

Possible Directions

  • New life domains — finances (budget tracking, bill management), health (medication, exercise, sleep)
  • Voice interface — deeper Home Assistant Assist integration for hands-free interaction
  • Calendar integration — sync with external calendars for deadline awareness and scheduling
  • Ambient reminders — Home Assistant automations (lights, display panels) for non-intrusive nudges
  • Multi-device sync — consistent experience across devices
  • Pattern learning — the system improves its friction and energy predictions over time based on activity log history

What's Speculative Here

All of it. Phase 4+ items are ideas that emerged from the design process and represent where the system could go. They will be scoped and prioritized based on what actually matters once Phases 1-3 are stable and in daily use.

How Phases Build on Each Other

Phase 1:  Data + API + Claude conversation
Phase 2:  + Scheduler + Notifications + Home Assistant
Phase 3:  + Web UI + Auth + Visual reporting
Phase 4+: + New domains + Voice + Calendar + Ambient

Each phase adds a capability layer without modifying the previous one. The API and data model from Phase 1 serve all future phases. The MCP contract from Phase 1 extends in Phase 2 with composite tools. The reporting from Phase 1 gets a visual representation in Phase 3.

Clone this wiki locally