Back to directory

OwnPilot

A self-hosted, privacy-first personal AI assistant platform with autonomous agents, multi-provider model routing, voice and browser automation, IoT device control, and Telegram and WhatsApp connectivity.

OwnPilot

OwnPilot routes requests across four native providers, OpenAI, Anthropic, Google and Zhipu, plus eight aggregator providers such as Together AI, Groq, Fireworks and OpenRouter, and any OpenAI-compatible endpoint, choosing by strategies such as cheapest, fastest, smartest, balanced or fallback. It also auto-discovers local models through Ollama, LM Studio, LocalAI and vLLM on the local network, and caches static Anthropic prompt blocks to cut input tokens on repeated requests.

Its more than 250 built-in tools span 33 categories, from files and code execution to email, media, git, translation, weather, finance and browser automation, but only four meta-tools are exposed to the model at once, with the rest reached through dynamic discovery and namespaced by their origin such as core, custom, plugin, skill or MCP. It supports soul agents and crews for delegated work, autonomous scheduled routines, an MCP client for external servers such as Filesystem, GitHub and Brave Search, and connects to Telegram and WhatsApp; all data stays on the user's own server.

More in Agent frameworks

See all 902
  • Recurse

    recurse.run

    A serverless harness that lets a coding agent design, test, and deploy its own specialist sub-agents as tools, MCP servers, or bots, defined by a schema-validated manifest.

  • ProofForge

    shulgin.is-a.dev

    An AI-agent pipeline that decomposes a mathematics problem, proves each piece, and formalizes the result in Lean 4 and Mathlib until the Lean kernel accepts it, with six of its proofs already merged into Google DeepMind's formal-conjectures repository.

  • local-coder

    github.com

    Builds a local coding-agent team on top of Ollama and OpenCode by detecting the machine's hardware, recommending models for seven specialist roles, and testing whether each assigned model can actually use tools, edit a repository and pass verification.