Case Auto-Formation

Talonic automatically groups related documents into cases by detecting shared entities like vendor names, contract IDs, and invoice numbers across extractions.

How cases form

When the four-phase pipeline extracts data from a document, it identifies entities — values linked to Field Registry entries marked as linkable. If two documents share the same vendor name or invoice number, Talonic clusters them into a single case. This inference-based clustering replaces manual document grouping.

Entity linking

Entity linking uses the Schema Graph to match values across schemas. A schema field marked as a linking entity (e.g., vendor_name) triggers cross-document matching. The 529-type ontology provides entity type definitions that guide the matching process.

Case resolution

Cases resolve automatically as new documents are extracted. Adding a delivery note that shares an order_id with an existing invoice case will merge the delivery note into that case. Each document in a case retains its own per-cell provenance trail.

Viewing cases

Cases are visible in the Talonic platform and can be queried through the search API. Use the filter endpoint to find all documents belonging to a specific case. Cases group documents across sources, enabling cross-source entity resolution.

Use cases

Cases are valuable for accounts payable (grouping invoices, purchase orders, and delivery notes for a vendor), compliance (linking contracts with amendments and addenda), and logistics (connecting bills of lading with customs declarations). The extraction data endpoint returns case membership for each extraction.