Self-hosted AI agents, chat interfaces and LLM tools
Self-hosted AI apps keep your agents, your chats and the keys they hold on a machine you control. An assistant that answers in Telegram or Slack needs a server that is always on, and the history stays where you decide.
OpenClaw, Hermes Agent and AstrBot are assistants you talk to in Telegram, Discord, Slack and other chat apps, and Octop serves a team or a family. Open WebUI is a chat interface for Ollama and OpenAI-compatible models, Flowise builds agents visually, and Langfuse traces and evaluates LLM apps. OpenConnector gives agents one MCP server for hundreds of services, and GitNexus turns a repository into a knowledge graph they can query.
- OpenClawAn AI assistant that lives in your chat apps
- Hermes AgentAn AI agent you talk to in Telegram, Discord or the browser
- Open WebUIA chat interface for OpenAI-compatible AI models
- AstrBotAn AI chatbot and agent for Telegram, Discord, Slack, QQ, WeChat and more, with plugins and a web dashboard
- FlowiseBuild AI agents and LLM workflows visually, then call them over an API or embed them as a chat widget
- GitNexusIndex a code repository into a knowledge graph and explore it, or let an agent query it over MCP
- LangfuseTracing, evals and prompt management for LLM apps
- OctopA self-hosted AI assistant for a team or a family: several users, expert agents, Telegram, Discord and more
- OpenConnectorOne MCP server and API for agents to act in GitHub, Gmail, Slack, Notion and hundreds of other services
- Open DesignDesign with AI: prototypes, slides and pages from a prompt, an open alternative to Claude Design
- OpenVikingA context database for AI agents: memory, resources and skills in one place
Missing an app? The templates are on GitHub, and anyone can add one with a pull request.