airweave: Retrieve Shared Context for AI Agents

Overview

Airweave is a retrieval layer between an organization’s data sources and its AI systems. It connects to apps, databases, and documents, then syncs and indexes their data so agents can search across sources through a unified interface.

It is intended to handle shared ingestion and retrieval infrastructure, including authentication and syncing, rather than requiring teams to build and maintain a separate pipeline for every integration or agent. Agents can access it through SDKs, a REST API, MCP, or integrations with agent frameworks.

Key Features

• Connectors for apps, databases, and documents, with the project describing support for 50+ integrations.
• Syncing and indexing to keep connected data available for retrieval.
• A unified search interface for retrieving context across multiple sources in one request.
• Python and TypeScript SDKs, plus a REST API.
• MCP and integrations with popular agent frameworks.
• A CLI for searching collections, managing sources, and triggering syncs.
• Self-hosted deployment with Docker Compose, with Kubernetes listed for production deployment.

Use Cases

• Agent developers can retrieve relevant context from several company tools without implementing a separate connector pipeline for each agent.
• Teams building internal assistants can search across sources such as Slack, Notion, Google Drive, and Jira when those sources are connected.
• Application developers can provide grounded search results to LLM features through the SDK or REST API.
• Operators can manage collections and trigger syncs from the CLI, or use its JSON output in scripts and agent workflows.

Getting Started

The self-hosted quickstart requires Docker and Docker Compose:

git clone https://github.com/airweave-ai/airweave.git
cd airweave
./start.sh

See the README for setup details and other ways to use Airweave.

Alternatives

• MaxKB: MaxKB builds knowledge-grounded agents and workflows, while Airweave focuses on syncing sources into reusable search infrastructure for agents.
• Guaardvark: Guaardvark is an all-in-one local AI studio with RAG, while Airweave specializes in connecting data sources to a unified search layer.

Considerations

• Self-hosting requires Docker and Docker Compose, and the README notes that the first startup can take a few minutes while services become healthy.
• The repository is marked as archived in the supplied GitHub metadata. Check its maintenance status and compatibility before adopting it for a new deployment.
• The system depends on multiple services. Its documented stack includes PostgreSQL, Vespa, Temporal, and Redis, which may add operational overhead compared with a standalone search library.