An overview of BeeAI core concepts and architecture
BeeAI is designed for a local-first experience - agents run directly on your laptop, giving you full control over your data and seamless integration with local inference engines like ollama.
The platform is built around several core components, illustrated in the diagram below:
An agent is a lightweight container that communicates using the Agent Communication Protocol (ACP). Agents may also define tools to extend their capabilities.
BeeAI comes with several built-in agents. Browse the full list in our Agent Catalog.
Dive into the beeai-server source code.
The BeeAI server coordinates the platform:
The server uses an official agent registry on GitHub to discover available agents.
BeeAI offers two main interfaces:
To start interacting with an agent: beeai run <name>
Or open the web interface at localhost:8333.
You can also integrate BeeAI into your Python applications using the ACP SDK. This makes it easy to coordinate multiple agents programmatically within your own workflows.
BeeAI supports observability through Arize Phoenix, an open-source tool for tracing agent behavior.
For more details, see our Observe Agents guide.