Omnisearch
Search across all documents, extractions, and field values using natural language queries. Omnisearch combines full-text search with Field Registry-aware semantic matching.
GET
/searchQuery parameters
| Parameter | Type | Description |
|---|---|---|
| q | string | Search query (natural language or field:value syntax) |
| type | string | Filter by result type: document, extraction, field |
| limit | integer | Max results (default 20) |
Example
curl "https://api.talonic.com/search?q=invoices%20from%20Acme%20over%201000&limit=10" \
-H "Authorization: Bearer tlnc_sk_live_7f3a...x9k2"How it works
Omnisearch parses natural language queries into field-level conditions using the Field Registry. The query "invoices from Acme over 1000" resolves to type=Invoice AND vendor_name=Acme AND total_amount>1000. Results include document metadata and relevance scores. For structured queries, use the filter endpoint.