Document upload: PDF, DOCX, TXT
Drag in contracts, playbooks, and support macros. Text is extracted, chunked, and embedded automatically so agents can cite the same sources your team trusts.
Your AI Works With Your Data. Not Hallucinations.
Upload documents, connect websites, get grounded answers.
Key capabilities
Website crawling and auto-indexing
Point at help centers, policies, or internal wikis. Crawls respect scope rules, dedupe URLs, and keep an index fresh without manual copy-paste.
pgvector semantic search
Meaning-based retrieval - not keyword games - so paraphrased questions still land on the right passages. All queryable from the same EU-ready Postgres footprint.
Automatic RAG in all agents
Templates pull from approved knowledge bases by default. Builders choose which corpora apply per agent so customer data never bleeds into internal experiments.
Source citations in responses
Answers link back to the chunk or page they came from. Reviewers can open the original, verify wording, and satisfy auditors without digging through logs.
How retrieval works under the hood
Vector embeddings
Documents and pages are embedded with models suited to your plan. Vectors live in Postgres with pgvector so similarity search stays fast and auditable next to the rest of your data.
Chunking strategy
Text is split into overlapping chunks with sensible defaults for prose vs. tables. Structure-aware splits keep headings near their sections so retrieval returns coherent context.
Refresh schedules
URL connectors re-crawl on a schedule you control. Uploads can be re-indexed when policies change. Stale sources are visible so owners know when to refresh.
SS-10
Knowledge base management - corpora, access, and health
1440 × 900pxSS-11
Document upload and ingestion status
1440 × 900pxExplore More Features
See how other platform capabilities work together.
Last updated: March 2026