# Honcho

Memory infrastructure for building stateful AI agents that understand people, other agents, groups and projects as they change over time, usable managed, local or self-hosted.

Honcho is memory infrastructure for building stateful agents that understand changing people, agents, groups, projects and ideas over time. An application stores messages and events, lets Honcho reason over them in the background, and then queries peer representations, session context, search results or natural-language insights from any model or framework.

It can be used managed at api.honcho.dev, run as a local stack with a single start command, or self-hosted as a FastAPI server, with client SDKs for Python and TypeScript and a companion CLI package. It offers reasoning-first memory that extracts conclusions from conversations rather than only matching text chunks, a peer-centric model that tracks users, agents, groups, projects and ideas as entities that change over time, and a multi-peer perspective modeling what one peer knows about another.

Documented integrations include giving a coding agent such as Claude Code, OpenCode, OpenClaw or Hermes persistent memory through MCP, and adding memory to a product directly through the Python or TypeScript SDK.

Website: https://honcho.dev/docs