AI-powered capture

Business value

Certain documents are too irregular for Document Capture’s existing recognition capabilities to capture reliably, such as documents with unusual layouts or non-standard fields. With LLM-based capture, a large language model will read the document and fill out the values automatically – even for barely legible documents or those in other languages.

Feature details

When you enable this AI capability and turn on the new AI-based capture setting on a template, Document Capture will extract the entire document – header and lines, or the header alone – in a single request to an AI model. This setting will work independently of the line recognition method, so you’ll be able to use it on templates both with and without line recognition.

Capture will run automatically when documents are imported. You can also run it on demand from the document card with your own prompt, for example to capture a non-standard field that's missing from the document.

Results will be stored within the document, so capturing it again will reuse them without initiating a new request.