Introduction
Reeboot is a personal AI agent that runs on your own machine and talks to you through the channels you already use — WhatsApp, Signal, or a browser-based chat. It has persistent memory, can search your documents, schedule tasks, and act proactively without being prompted.
Who It’s For
Reeboot is for people who want an AI assistant that is:
- Private — your conversations, your documents, and your memory stay on your machine
- Always available — connected to WhatsApp or Signal, not locked inside a browser tab
- Yours to extend — add tools, connect services, write skills in plain Markdown
It works equally well as a simple chat assistant and as a capable autonomous agent that can run scheduled jobs, monitor things, and take action on your behalf.
What Makes It Different
| Reeboot | Typical AI chat | |
|---|---|---|
| Runs locally | ✅ Single Node.js process | ❌ Cloud-only |
| Talks on WhatsApp / Signal | ✅ Native integration | ❌ Web app only |
| Remembers you across sessions | ✅ Persistent memory files | ❌ Stateless |
| Searches your documents | ✅ Local vector embeddings | ❌ No document access |
| Schedules its own tasks | ✅ Cron-based scheduler | ❌ No proactive actions |
| Fully configurable | ✅ JSON config, TypeScript extensions | ❌ No customisation |
| Your choice of LLM | ✅ 8 providers including Ollama | ❌ Single provider |
How It Works
Reeboot is a single process. All your channels (WhatsApp, Signal, WebChat) connect to one orchestrator, which routes each message to the AI agent and returns the response to the right place.
WhatsApp ──┐
Signal ──┼──► Orchestrator ──► AgentRunner (pi) ──► LLM
WebChat ──┘ │
└──► Scheduler (background tasks)
State is stored in a single SQLite database at ~/.reeboot/db/reeboot.db. Memory lives in two Markdown files at ~/.reeboot/agent/. Everything else is in ~/.reeboot/config.json.
Next Steps
- Installation — install reeboot and run it for the first time
- Quick Start — be talking to your agent in five steps
- Setup Wizard — a detailed walkthrough of the first-run wizard