A local-first daemon that watches your tools, builds a living world model, and surfaces discoveries you didn't know you needed. Desktop app, CLI, web dashboard, Docker, or MCP.
A quiet analyst that speaks first only when something matters.
Entities, relationships, events, patterns, and situations — all tracked in a local SQLite database that grows with every data source interaction.
Native app for Windows, macOS, and Linux. System tray with live score, global hotkey (Ctrl+Shift+O), native OS notifications, and a setup wizard.
Query your world model in natural language. "What's happening with Acme?" OWL searches entities, events, discoveries, and patterns to answer.
Interactive D3.js knowledge graph, live discovery feed, OWL Score gauge, entity detail sidebar, source filters, and event timeline. Auto-refreshes every 60s.
Related discoveries are linked into narrative threads. When enough accumulate, OWL generates meta-discoveries — insights about your insights.
Statistical baselines per source, day, and hour. Z-score detection flags volume spikes and unexpected silence before you notice.
Model Context Protocol server for Claude Desktop, Cursor, and Windsurf. Ask "What's happening in my world?" from any MCP-compatible AI client.
Relationship graph with path finding, community detection, bridge entity identification, and hub ranking. See your world as a network.
React to discoveries via Telegram, Slack, or CLI. OWL learns what you value and adapts its filters, confidence calibration, and LLM prompts.
One-command deployment with docker compose up. Runs daemon + dashboard, persistent volumes, healthchecks, and Ollama integration built-in.
Every Sunday, a narrative summary: top discoveries, new entities, active situations, and what to watch for next week.
Non-urgent discoveries are held during your configured quiet period. Urgent ones still break through (unless you mute them too).
OWL lives in your system tray. Always watching. One hotkey away.
Press Ctrl+Shift+O from any app to summon OWL instantly. Press again to hide.
New discoveries push to your OS notification center. Click to jump straight to the insight.
OWL Score, daemon status, recent discoveries, and quick actions — all in the system tray.
Sleek, modern look with custom titlebar. Remembers your position and size across restarts.
4-step guided setup on first launch: LLM connection, data sources, channels, and preferences.
Toggle in tray menu. OWL starts with your computer and watches silently in the background.
Natural language queries against your entire world model.
OWL searches your entities, events, discoveries, patterns, and situations to answer questions. It understands entity names, detects fuzzy matches, and pulls deep relationship context automatically.
Use owl ask --chat for multi-turn interactive conversations with your world model, complete with follow-up suggestions.
Choose what fits your workflow. Mix and match.
Native app with tray, hotkey, and notifications. Windows, macOS, Linux.
npm run electron:dev
The original. Runs in background with cron scheduling.
owl start
Interactive knowledge graph with D3.js. Real-time discovery feed.
owl dashboard
One-command deploy. Daemon + dashboard in a container.
docker compose up -d
Query OWL from Claude Desktop, Cursor, or Windsurf.
Add to mcp config
Plugins poll your data sources
Extract entities & relationships
LLM finds patterns & insights
Novelty, confidence, importance
To the channels you live in
Data sources in. Discoveries out. Through the tools you already use.
Add OWL as an MCP server and query your world model from Claude Desktop, Cursor, or any MCP-compatible client.
{
"mcpServers": {
"owl": {
"command": "node",
"args": ["/path/to/owl/src/mcp/server.js"]
}
}
}
Your data never leaves your machine. OWL runs as a local daemon with a local SQLite database. The only external calls are to your data source APIs and your configured LLM.
owl forget deletes instantlyowl export for full data portabilityInstall OWL in 60 seconds. See it work instantly with owl demo.