什么是Airweave？开源上下文检索层详解

What is Airweave?

Airweave connects to your applications, tools, and databases, continuously synchronizes their data, and exposes it through a unified, large language model (LLM)-friendly search interface. AI agents query Airweave to retrieve relevant, grounded, and up-to-date context from multiple sources in a single request.

Airweave 连接到您的应用程序、工具和数据库，持续同步其数据，并通过一个统一的、对大语言模型（LLM）友好的搜索接口将其暴露出来。AI 代理通过查询 Airweave，可以在单个请求中从多个来源检索相关的、有依据的、最新的上下文信息。

Where It Fits in Your Architecture

Airweave sits between your data sources and AI systems as shared retrieval infrastructure. It handles authentication, ingestion, synchronization, indexing, and retrieval, so you don't have to rebuild fragile data pipelines for every new agent or integration.

Airweave 作为共享的检索基础设施，位于您的数据源和 AI 系统之间。它处理身份验证、数据摄取、同步、索引和检索，因此您无需为每个新的代理或集成重建脆弱的数据管道。

How It Works

The Airweave system operates through a streamlined, three-step process:

Connect your data sources: Integrate your applications, databases, and documents using one of 50+ supported connectors.

连接您的数据源：使用 50 多个受支持的连接器之一，集成您的应用程序、数据库和文档。
Sync and index: Airweave automatically synchronizes, processes, and indexes your data, creating a unified retrieval layer.

同步与索引：Airweave 自动同步、处理和索引您的数据，创建一个统一的检索层。
Query and retrieve: Agents query Airweave via SDKs, REST API, MCP, or native integrations with popular agent frameworks to retrieve relevant, grounded context on demand.

查询与检索：代理通过 SDK、REST API、MCP 或与流行代理框架的原生集成来查询 Airweave，按需检索相关的、有依据的上下文。

Quickstart Guide

Cloud-Hosted Option

The fastest way to get started is via our managed cloud service: app.airweave.ai

最快开始的方式是通过我们的托管云服务：app.airweave.ai

Self-Hosted Deployment

For full control, you can deploy Airweave in your own environment. Ensure you have Docker and Docker Compose installed, then run the following commands:

为了完全控制，您可以在自己的环境中部署 Airweave。请确保已安装 Docker 和 Docker Compose，然后运行以下命令：

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

Once complete, access the interface at http://localhost:8080.

完成后，通过 http://localhost:8080 访问界面。

SDK Integration

Airweave provides official SDKs for popular programming languages to simplify integration into your agent workflows.

Airweave 为流行的编程语言提供官方 SDK，以简化集成到您的代理工作流程中。

Python

pip install airweave-sdk

from airweave import AirweaveSDK

client = AirweaveSDK(api_key="YOUR_API_KEY")
results = client.collections.search(
    readable_id="my-collection",
    query="Find recent failed payments"
)

TypeScript/Node.js

npm install @airweave/sdk

Technology Stack

Airweave is built on a modern, robust, and scalable technology stack:

Frontend: React with TypeScript, utilizing ShadCN UI components.

前端：使用 TypeScript 的 React，采用 ShadCN UI 组件。
Backend: FastAPI (Python) for high-performance API services.

后端：使用 FastAPI（Python）提供高性能 API 服务。
Databases: PostgreSQL for metadata storage and Vespa for vector search and retrieval.

数据库：PostgreSQL 用于元数据存储，Vespa 用于向量搜索和检索。
Workers & Orchestration: Temporal for workflow orchestration and Redis for pub/sub messaging.

工作流与编排：Temporal 用于工作流编排，Redis 用于发布/订阅消息传递。
Deployment: Docker Compose for local development and Kubernetes for production deployments.

部署：Docker Compose 用于本地开发，Kubernetes 用于生产环境部署。

Contributing and License

We are an open-source project and welcome contributions from the community. Please see our Contributing Guide for details on how to get involved.

我们是一个开源项目，欢迎社区的贡献。有关如何参与的详细信息，请参阅我们的贡献指南。

Airweave is released under the MIT License.

Airweave 基于 MIT 许可证 发布。

Connect with Us

Discord · Issues · Twitter

联系我们

Discord · Issues · Twitter