OpenHuman
Open-source, local-first personal AI with persistent memory and its own fleet of automated agents.
🔗 Visit OpenHuman
Description
Most AI assistants forget everything the moment you close the chat, and every automation you set up lives inside someone else's cloud service. OpenHuman is built around the opposite idea: an AI that remembers things about you over time and runs its automations on your own machine.
OpenHuman is an open-source, local-first personal AI platform that combines persistent memory (stored locally in SQLite), an orchestration engine for running fleets of autonomous agents, and managed tools for web search, scraping, browser automation, and voice or image processing. It connects to over 100 OAuth services and thousands of MCP servers, and routes messages across 17 channels including email, Slack, Discord, and Teams. Automation runs through approval-gated workflows with replayable run journals, aimed at technical users who want serious agent automation without handing their data and workflows entirely to a third-party cloud platform.
💬 Our review
The short version: if you want an AI assistant that actually remembers context across sessions and can run real automations — not just answer questions — while keeping your data and API keys under your own control, OpenHuman is a genuinely ambitious open-source attempt at that.
It's not really competing with consumer chat assistants; the more honest comparison is against other emerging agent-orchestration platforms, and OpenHuman's differentiator is the combination of local-first storage, bring-your-own-model flexibility (your own API keys or local Ollama models), and breadth of integrations (100+ OAuth services, thousands of MCP servers) in one open-source package. The tradeoff is exactly what you'd expect from that ambition: it's a self-hosted, technical project, not a polished consumer app, so it demands real setup effort and comfort with agent/automation concepts. For a developer who wants full control over a personal AI's memory and integrations, it's worth the setup time; for someone who just wants a chatbot, it's overkill.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Modèle par abonnement, ou clés API/modèles Ollama locaux apportés par l'utilisateur pour éviter tout coût récurrent
Pros
mémoire persistante stockée localement, pas dans le cloud d'un tiers
orchestration de flottes d'agents autonomes avec validation humaine
plus de 100 intégrations OAuth et des milliers de serveurs MCP
compatible clés API personnelles ou modèles Ollama locaux, pas de lock-in fournisseur
Cons
projet technique en self-hosting, pas une app grand public clé-en-main
courbe de mise en place significative
peu de recul/retours indépendants disponibles à ce stade