> ## Documentation Index
> Fetch the complete documentation index at: https://docs.neudocs.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Structured capture

> Extract firm-defined names, dates, totals, and other fields during document evaluation.

Structured capture asks the correctness evaluation to return specific values from a document, such as a tax year, legal name, statement date, or total.

## Configure capture fields

Add concise field names to a template file item. Use terms that appear in the document and include context when necessary:

* `Tax year`
* `Account holder name`
* `Statement ending date`
* `Gross income total`

Avoid vague fields such as `amount` when several amounts may appear.

## Review captured values

Captured values appear with the correctness result in the review workflow. Compare them to the source document before copying them into accounting, tax, or legal systems.

## Cost behavior

When the optional AI judge runs, correctness and structured capture share that evaluation rather than making a separate model call solely for extraction.

## Limitations

Extraction quality depends on scan quality, document layout, language, provider capability, and field wording. Unsupported or unreadable documents may return no values. Captured data is review assistance, not authoritative accounting data.


## Related topics

- [Quickstart](/quickstart.md)
- [Evidence ledger](/verification/evidence-ledger.md)
- [Request templates](/workflow/templates.md)
- [Settings and branding](/workspace/settings.md)
- [Security and privacy](/verification/security-privacy.md)