Give your agents
knowledge.

File uploads

PDF, DOCX, CSV, and XLSX files. Small spreadsheets are injected directly; larger files are chunked and indexed.

Websites and webpages

Crawl an entire site or add individual page URLs. Content is extracted and indexed automatically.

Live connector sources

Google Docs, Notion pages, and Confluence — agents read the latest version in real time.

Content-aware chunking

Documents are split intelligently — spreadsheets by rows (with headers preserved), text by semantic boundaries.

Vector embeddings

Each chunk is embedded with OpenAI's text-embedding-3-small for fast, accurate semantic search.

Hybrid search

Agents find the right content even when customers don't use exact terms from your documents.

Always up to date

Agents read the latest version of your Google Docs, Notion pages, and Confluence spaces in real time.

No re-uploading

Update a Google Doc and your agent immediately has the latest content. No manual sync needed.

Google Picker integration

Select Google Docs and Sheets using the native Google Picker — no manual copy-pasting of document IDs.

All documents or selected

Give an agent access to everything in a source, or hand-pick specific documents.

Per-agent granularity

The billing agent sees billing policies. The support agent sees the full knowledge base. Different agents, different access.

Inline scope toggle

Switch between "All documents" and "Selected" right on the source row — no extra modals or settings pages.