AI that knows
your documents,
not just the world.
Build private, queryable knowledge collections from your research documents and data. Any AI tool can plug in — through a secure, scoped API.
How it works
From documents
to answers.
Upload your content once. Query it forever — through a clean API or directly in the catalogue.
Ingest anything
Upload PDFs, connect Zotero libraries, or pipe in structured data. Fetchlake extracts, chunks, and embeds your content automatically.
Semantic retrieval
Dense vector search combined with BM25 sparse retrieval finds the right passages — not just keyword matches.
API-first
Every collection is instantly queryable via a clean REST API. Connect to any LLM, workflow, or application in minutes.
Private and scoped
Your knowledge stays yours. API keys are scoped, rate-limited, and revocable — with full audit trails.
Query your knowledge
in minutes
Connect your knowledge collections to applications, internal assistants, and AI systems with a simple API call. Switch between research, policy, and operational use cases to see grounded answers returned from your documents.
- Works across research, policy, and internal operationsConnect any document collection to any application or assistant.
- Returns answers grounded in your filesResponses cite specific documents with section and page references.
- Supports API and MCP integrationsUse REST APIs or MCP to connect to internal tools, bots, and workflows.
See how one API works across different knowledge collections
import requests
response = requests.post(
"https://api.fetchlake.ai/v1/query",
headers={
"Authorization": "Bearer FETCHLAKE_API_KEY",
"Content-Type": "application/json",
},
json={
"collection": "mtss-research",
"query": "What interventions improve school attendance?",
"top_k": 5
}
)
print(response.json())Grounded answers returned from your knowledge collection.
Ready to get started?
Start free and build your first knowledge collection today. No credit card required.