OpenHuman
OpenHuman is a free, open-source personal AI for Mac, Windows and Linux that keeps data local and orchestrates fleets of agents. Built in Rust, it is positioned as a fast, efficient open-source agent harness.
Visit tinyhumansai/openhumanOverview
OpenHuman is a free, open source personal AI for Mac, Windows and Linux. It keeps data local, builds persistent memory from connected apps, and uses an orchestration engine to manage fleets of agents. Its core is written in Rust and supports desktop, browser, and terminal clients.
Key Features
- Memory engine decides what to keep, compress, and surface when an agent needs context; it can mirror memory as an Obsidian vault.
- Connects apps through OAuth, MCPs, and Skills, with support for up to 100 apps, 10,000 MCPs, and 90,000 Skills.
- Workflows engine orchestrates agents for simple or complex automation graphs; the agent proposes workflows for review and save.
- Runs fleets of agents in one Rust core process, with measured low marginal cost per additional agent.
- Pluggable engines for LLM, embeddings, memory, and web search, including local models and bring-your-own-key providers.
- Ships as a desktop app for Windows, macOS, and Linux, plus a browser UI and terminal client.
Use Cases
- Use as a personal AI that lives on your laptop and learns from your connected work apps.
- Orchestrate multiple agents to carry out tasks across tools instead of only replying in chat.
- Automate workflows with scheduled, app-event, or manual triggers and approval gates.
- Build a second brain from connected apps with a local memory graph.
Getting Started
- Download installers from the official website or the GitHub Releases page.
- For terminal installs, use Homebrew, Debian/Ubuntu .deb, AUR, install scripts, and platform notes; see INSTALL.md.
- The desktop app runs on Windows, macOS, and Linux; browser UI and terminal client use the same core.
Deployment & Requirements
- Desktop app supports Windows, macOS, and Linux.
- The same core can run as a browser SPA or a terminal client.
Before You Adopt
- License: GPL-3.0. Review its terms before using, modifying, or distributing the project.
- The project is labeled an early beta and under active development.
- Remote memory engine adapters ship, but a live install cannot yet switch to one through the host binding.
- The Jev decision model depends on a TinyHumans credential for full behavior and falls back to BM25 without one.