Knowledge engine
RAG — trained on your data
Your agent knows your website, FAQs, PDFs and product data, and answers with source attribution.
Details
How it works
We index your content using semantic search. For every request the agent fetches the most relevant documents and grounds its answer on them. Result: current, accurate answers, with a link back to the source.
- Supports PDF, Markdown, HTML, Notion, Google Drive, Confluence
- Re-indexing automatically on changes or via webhook
- Multilingual embeddings — content in DE, queries in EN work
- Hybrid search: keyword + semantic for higher quality
- Source attribution on every answer: the editor sees where the agent quoted from
02by-the-numbers
PerformanceRAG by the numbers
- Answer accuracy
- 92%
- Latency
- <400ms
- Documents per agent
- 10k+
- Hosting
- EU
03faq
Frequent questions.
Ready for a live demo?
We'll show you how this looks for your use case — 30 minutes, no strings attached.