Airweave Block
Search and retrieve data from Airweave collections
The Airweave block brings semantic search into your workflows. Query a collection with natural language, choose a retrieval strategy, and optionally generate an AI-powered answer from the matched results. Reach for it whenever a workflow needs to pull relevant context out of an Airweave collection.
Overview
| Property | Value |
|---|---|
| Type | airweave |
| Category | tools |
| Color | #6366F1 |
When to Use
- Retrieve relevant documents or records from an Airweave collection using natural-language queries.
- Power RAG-style flows by feeding search results into a downstream agent or prompt.
- Get a synthesized, AI-generated answer directly from a collection without a separate LLM step.
- Tune retrieval quality by switching between semantic, keyword, hybrid, or auto strategies.
- Improve precision on large result sets by enabling reranking.
- Surface context from synced sources (GitHub, Slack, and others) inside a multi-step workflow.
Configuration
Collection ID
The readable ID of the Airweave collection to search. Required. Can be passed from another block with {{blockName.collectionId}}.
Query
The natural-language search query describing what you want to find. Required. Supports {{blockName.field}} references to inject dynamic queries.
Result Limit
Maximum number of results to return. Dropdown — choose one of:
| Label | ID |
|---|---|
| 5 | 5 |
| 10 | 10 |
| 20 | 20 |
| 50 | 50 |
Retrieval Strategy
How matches are scored and retrieved. Dropdown — choose one of:
| Label | ID |
|---|---|
| Auto | auto |
| Semantic | semantic |
| Keyword | keyword |
| Hybrid | hybrid |
hybrid combines vector and keyword scoring and is the default strategy used by the underlying tool when no value is set.
Generate Answer
Toggle (switch). When enabled, Airweave synthesizes an AI-generated natural-language answer from the search results and returns it in the completion output.
Rerank Results
Toggle (switch). When enabled, the returned results are reordered by an LLM for higher relevance precision.
API Key
Your Airweave API key, stored securely as a password field. Required. Use {{AIRWEAVE_API_KEY}} to reference an environment variable.
Inputs & Outputs
-
Inputs:
collectionId(string) — Collection ID to searchquery(string) — Search query textlimit(string) — Maximum number of results to returnretrievalStrategy(string) — Retrieval strategy (auto,semantic,keyword, orhybrid)generateAnswer(boolean) — Whether to generate an AI-powered answer from the resultsrerank(boolean) — Whether to rerank results for improved relevanceapiKey(string) — Airweave API key for authentication
-
Outputs:
results(json) — Array of search result objects, each containing:entity_id,source_name,md_content(markdown content, optional),score(relevance score),metadata(optional),breadcrumbs(navigation path, optional),url(link to original content, optional)completion(string) — AI-generated answer to the query; only present whengenerateAnsweris enabled
Tools
Airweave Search (airweave_search) — Sends a POST request to https://api.airweave.ai/collections/{collectionId}/search using the provided API key. Supports semantic, keyword, and hybrid retrieval strategies. Optionally expands the query for better recall, reranks results with an LLM, and synthesizes a natural-language answer. Returns an array of result objects and an optional completion string.
YAML Example
airweave_1:
type: airweave
name: "Airweave"
inputs:
collectionId: "my-collection-id"
query: "{{start.input}}"
limit: "10"
retrievalStrategy: "hybrid"
generateAnswer: true
rerank: false
apiKey: "{{AIRWEAVE_API_KEY}}"
connections:
outgoing:
- target: next-block-id