Runs unified search across both unstructured documents and structured connector-backed data in your environment. Memic classifies the incoming query and intelligently routes it — natural-language questions are answered by semantic search over indexed documents, while structured queries (filters, aggregations, lookups) hit connector-backed sources. Returns ranked results with source citations plus routing metadata so you can see which path was taken.
Memic API key. Every key is bound to exactly one environment — get one from the Memic dashboard under API Keys.
Request model for semantic search.
The search query text
1 - 1000"What are the key findings in the cybersecurity report?"
Optional project ID to limit search scope
Optional list of file IDs to limit search scope
Optional environment slug to filter search ('staging' or 'production')
Number of results to return (max 50)
1 <= x <= 50Minimum similarity score threshold (0.0 to 1.0)
0 <= x <= 1Whether to include file and chunk metadata in results
Optional: Force query to use a specific connector
Optional: Force query routing (structured, semantic, or hybrid)
structured, semantic, hybrid Metadata filters for vector search (reference_id, page_number, etc.). Used in lookup-based search where SQL returns filter values.
Successful Response
Response model for semantic search.
Original search query
Total number of results returned
Search execution time in milliseconds
Organization namespace used for search
Filters applied to the search
All search results by type
Query routing information