Context

Most AI assistants are built for the companies that made them, not for the person using them. Your conversations live on someone else's server. Your context resets every session. You have no visibility into what the agent actually did, when, or why. And you can only reach it when you open a browser tab.

We wanted the opposite: an agent that runs on hardware you control, remembers what matters, meets you where you already are, and keeps a complete, inspectable record of every action. An agent built for the user, not the platform.

What We Built

reeboot is a personal AI agent installable via npm and runnable locally or on a private server. It connects to the channels you already use — WhatsApp, Signal, and web chat — with a clean API for adding custom channel adapters. Every design decision prioritises sovereignty and durability over convenience:

• Memory — facts and preferences persist across sessions, with sleep consolidation that distils what matters after every day
• Local Knowledge Base — vector embeddings over your own documents, fast and fully offline
• Scheduler — cron jobs, intervals, and one-shot tasks so the agent acts without being asked
• Observability — a structured audit log of every action, timestamped and inspectable
• Sandbox — OS-level confinement on macOS and Linux, with Docker deployment fully supported
• MCP + Skills — native MCP server support and plain-Markdown skills; connect 16,000+ tools or write your own without code
• Budget Management — token spending limits per task, per day, or per model
• Resilience — crash recovery, outage detection, scheduler catchup, and wake-up notifications

The core stays deliberately small — no bloat, no everything-framework. Every feature earns its place. And the entire project was planned and built using our own reespec methodology.

Services Proven

reeboot demonstrates our depth in AI capabilities and agentic systems engineering. We don't just integrate AI tools; we design and build autonomous agents with real-world constraints: persistence, auditability, multi-channel integration, local inference, and resilient scheduling. If your organisation needs a custom AI agent — private, embedded in your workflow, and built to last — this is the work we do.