canonical: https://jentic.com/apis/airweave.ai/airweave

# Airweave API

Jentic publishes the only available OpenAPI specification for Airweave API, keeping it validated and agent-ready. Airweave is a search and data-management platform that lets developers build retrieval over heterogeneous SaaS sources (Notion, Slack, Google Drive, Stripe, and more). The 27-endpoint API exposes collections (search-ready indexes), source connections (sync configurations), webhook ingestion, and three search modes - instant keyword search, classic semantic search, and agentic search with optional streaming. Authentication uses an x-api-key header.

## For AI agents

Index data from SaaS sources into searchable collections, then run instant, classic, or agentic search to retrieve relevant content for an AI agent. Streaming agentic search is supported.

## Scope

Does not host LLMs, generate embeddings as a standalone service, or run web crawls of arbitrary sites - use for SaaS-source ingestion and search only.

## Capabilities

- Create and manage collections that index data from connected SaaS sources
- Configure source connections to sync from Notion, Slack, Google Drive, Stripe, and other platforms
- Run instant keyword search across a collection for low-latency lookups
- Run classic semantic search for embedding-based retrieval
- Run agentic search that plans multi-step retrieval and optionally streams the response
- Trigger and cancel sync jobs and inspect job history per source connection
- Receive webhook events for real-time ingestion of upstream changes

## Use cases

### Unified RAG over SaaS data

Build a retrieval-augmented application that searches across Notion, Slack, Google Drive, and other tools without writing per-source connectors. Airweave handles ingestion, embedding, and indexing into collections, and exposes search via instant, classic, and agentic modes.

Example prompt: Create a collection 'company-knowledge', attach a Notion source connection, and run a classic search for 'Q4 roadmap commitments'

### Agentic enterprise search

Use the agentic search endpoint to let an AI agent plan a multi-step retrieval - for example, decomposing 'find the latest customer feedback on the new pricing page' into sub-queries across docs and chat. The streaming variant returns intermediate reasoning so a UI can render progress.

Example prompt: POST to /collections/{readable_id}/search/agentic/stream with the query 'recent customer complaints about pricing' and consume the stream

### Sync orchestration and observability

Manage source connections programmatically - create, update, run, and cancel sync jobs, and list job history to surface failures. Useful when an agent operates a knowledge platform and must respond to ingestion incidents.

Example prompt: List sync jobs for source-connection {id}, identify the most recent failed job, and trigger a new run via POST /source-connections/{id}/run

### AI agent integration via Jentic

Wire Airweave into a Jentic-powered agent so it can search company knowledge by intent. The agent loads the Airweave operation schema at runtime, your Jentic One instance holds the API key, and the agent calls instant, classic, or agentic search depending on the latency and depth required.

Example prompt: Search Jentic for 'agentic search across SaaS data', load the matching Airweave operation, and execute against the user's collection

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | /collections | Create a new collection |
| POST | /collections/{readable_id}/search/instant | Run instant keyword search |
| POST | /collections/{readable_id}/search/classic | Run classic semantic search |
| POST | /collections/{readable_id}/search/agentic | Run agentic multi-step search |
| POST | /collections/{readable_id}/search/agentic/stream | Stream agentic search results |
| POST | /source-connections | Create a source connection |
| POST | /source-connections/{source_connection_id}/run | Trigger a sync job |

## Key resources

- **Collections** — Create and manage searchable indexes that aggregate data from connected sources
- **Source Connections** — Configure and run sync jobs that pull data from Notion, Slack, Drive, Stripe, and other sources
- **Sources** — List available source types and inspect their schemas
- **Search** — Three modes - instant, classic, agentic - plus a streaming variant of agentic
- **Webhooks** — Receive real-time ingestion events from upstream sources

## Why Jentic

- **Setup:** Wiring Airweave by hand means handling its x-api-key header, choosing between instant, classic, and agentic search modes, and writing your own retry logic. Through Jentic you install once, import Airweave from the API Directory, store the key once, and your agent calls it.
- **Permission scoping:** Airweave puts the collection id in the URL path (/collections/{readable_id}/search/instant), so a rule can pin your agent to one collection. You choose the operations it may call, so running a source connection is only included if you add it.
- **Credential handling:** Your Airweave key is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'search across my SaaS data', and Jentic returns the matching Airweave search operation with its input schema so the agent calls the correct search mode without browsing the reference docs.

## Related APIs

- **Pinecone** — Pure vector database - you bring the connectors and embedding pipeline.
- **Notion API** — Connect Notion as an Airweave source to make pages searchable.
- **OpenAI API** — Generate the answer or summary from Airweave's retrieved context.

## FAQ

### Why is there no official OpenAPI spec for Airweave API?

Airweave does not publish an OpenAPI specification. Jentic generates and maintains this spec so that AI agents and developers can call Airweave API via structured tooling. It is validated against the live API and kept up to date. Get started with Jentic One, the self-hosted execution layer.

### What authentication does the Airweave API use?

Airweave uses an API key in the x-api-key header. Through Jentic, the API key is stored in the encrypted Jentic One instance and injected at execution time, so the agent never sees the raw key.

### What is the difference between instant, classic, and agentic search?

Instant search at /collections/{id}/search/instant is keyword-based and optimised for low latency. Classic search at /search/classic runs semantic embedding retrieval. Agentic search at /search/agentic plans multi-step queries and can be streamed via /search/agentic/stream for live UIs.

### Can I run agentic search through Jentic?

Yes. Use the Jentic search query 'agentic search across collected data'. Jentic returns the POST /collections/{readable_id}/search/agentic operation with its input schema so the agent can pass the query and collection id directly.

### How do I sync a new data source into Airweave?

Create a source connection via POST /source-connections with the source short name and credentials, then trigger a sync via POST /source-connections/{id}/run. Track progress with GET /source-connections/{id}/jobs.

### Are webhooks supported for real-time ingestion?

Yes. The /webhooks resource exposes message listing endpoints used to inspect inbound events. Configure upstream sources to post to your Airweave webhook URL so changes propagate without waiting for a scheduled sync.

### Can I limit what my agent is allowed to do with the Airweave API?

Yes. Because Jentic One is self-hosted, your own rules decide which Airweave operations and credentials the agent may use. Since the collection id sits in the URL path, such as /collections/{readable_id}/search/instant, you can pin the agent to a single collection and to read-only search modes like instant, classic, or agentic. Write operations such as running a source connection via /source-connections/{id}/run are included only if you add them.