Semantic graph search
Last updated 2026-06-14
Definition
Semantic graph search lets Quri chat find the right emails, chat summaries and recommendations by meaning, not exact words. It turns your question into a vector, compares it against everything Quri has written down, and returns the closest records — so "what did we tell users about the checkout bug?" surfaces the message even if it never said "checkout".
How to do this in Quri
- Connect your sources so Quri keeps writing emails, chat summaries and recommendations into the graph.
- Ask a meaning-based question in /app/chat, like "find the notes about slow checkout".
- Quri embeds your question and ranks the stored records by how close they are in meaning.
- Open the records it cites to read the exact emails or recommendations behind the answer.
Frequently asked
- How is this different from searching for a keyword?
- It matches on meaning, so a search for "checkout bug" still finds a message that said "payment failed at the cart" — the words differ but the intent is the same.
- Where do the embeddings come from?
- Quri embeds your records as it ingests them, using a local model by default so nothing leaves your environment and there is no per-word cost. Only your live question is embedded at the moment you ask.