Your personal AI-powered task manager — self-hosted, private, and built to actually help you get things done.
No accounts. No SaaS. No data leaving your machine unless you choose it.
Features • Setup • Screenshots • Releases • Tech Stack
Dooable is a self-hosted task manager that runs entirely on your machine. It connects to a local AI model (via LM Studio) or Anthropic Claude to give you intelligent planning assistance — without sending your data to any third-party service.
It's not a simple to-do list. It's a productivity layer that learns your patterns, warns you when you're overloaded, and generates a daily briefing so you always know what matters most.
AI that actually knows your workload — Dooable analyzes your 30-day completion history and flags when today's task list exceeds what you can realistically finish. It doesn't just track tasks — it helps you plan honestly.
A daily briefing, every morning — Instead of opening a list and deciding what matters, Dooable generates a prioritized briefing: what to do now, what to keep on your radar, and a plain-English summary of your day. Works with LM Studio (local, free) or Anthropic Claude.
Type tasks in plain English — "Submit quarterly report Friday at noon, high priority" — Dooable parses the date, time, and priority automatically. No dropdowns, no fuss.
Everything in one place — Tasks, subtasks, tags, notes, file attachments, recurring schedules, Kanban board, calendar, backlog, projects, templates — all connected and all yours.
Understands your patterns — Passive intelligence detects when you keep deferring the same tasks, spots momentum drops before they become problems, and surfaces procrastination signals in your briefing.
Privacy-first by design — Your data lives in a Postgres database on your own machine. The local AI runs entirely offline. Anthropic Claude is opt-in. Nothing is required to leave your network.
| AI daily briefing | Prioritized morning summary — do-now, keep-on-radar, plain-English overview |
| Capacity analysis | 30-day completion history + real-time overhang alerts |
| NLP quick-add | Parse tasks from natural language (date, time, priority, notes) |
| Smart reschedule | AI suggests new dates for overdue tasks based on your workload |
| Kanban + Calendar | Drag-drop board and 2-week calendar with AI-assisted rebalancing |
| Focus Mode | Keyboard-driven single-task timer with elapsed vs. estimated tracking |
| Drift detection | Audit trail of rescheduled, deferred, and dropped tasks |
| Weekly & monthly reviews | AI-generated retrospectives on your actual completion patterns |
| Insights dashboard | Tag breakdown, priority drift, planning accuracy, AI narrative |
| Projects & templates | Organize tasks into projects, save reusable blueprints |
| Dependency graph | Visual map of blockers and critical paths |
| Command palette | Cmd+K search and navigation across all tasks and pages |
1. Create .env
POSTGRES_DB=dooable
POSTGRES_USER=dooable
POSTGRES_PASSWORD=changeme
API_KEY=your-secret-token-here
ANTHROPIC_API_KEY=
GROQ_API_KEY=No GPU? Use Groq (free) — Dooable runs its AI features on a local LM Studio model by default, but you can use Groq's free cloud API instead:
- Sign up at console.groq.com (free, no credit card).
- Go to API Keys → Create API Key and copy the
gsk_…key. - Paste it into
GROQ_API_KEY=in your.env(or later in Settings → AI Model → Groq inside the app). - After login, open Settings → AI Model and select Groq.
The free tier is rate-limited but plenty for personal task management. Embeddings and screenshot import still require LM Studio.
2. Download compose file
curl -O https://raw.githubusercontent.com/lundsonn/dooable/main/docker-compose.prod.yml3. Start
docker compose -f docker-compose.prod.yml up -dOpen http://localhost:3000.
→ Full setup instructions, LM Studio config, and security notes in docs/setup.md.
v1.2.0 — Time-of-day completion heatmap, LLM task classification, peak pattern insights.