Skip to main content
Once a workflow has been created and documents have been processed, the next step is to review and validate the extracted data. This section is about:
  • Reviewing results
  • Validating data
  • Correcting outputs
  • Preparing data for export or use

Important distinction

Before going further, it’s important to clarify a common source of confusion.

Refining a workflow

During creation - improves how data is extracted in the future.

Validating results

After processing - confirms whether extracted data is correct for specific documents.
You can always go back and refine the workflow if results are not good enough.
anyformat will learn from validation and corrections over time to improve future extractions.
This section focuses on validating results, not on editing prompts.

Table view

Reviewing data at scale

The table view lets you review all extracted data across documents. You’ll see:
  • One row per document
  • One column per field in the schema
  • Metadata such as creation date, processed date, and validator

Table actions

From the table view, you can:
  • Run the workflow on unprocessed documents
  • Download processed results (CSV, Excel, JSON)
  • Delete selected documents
  • Open a document in page view
  • Download results per document (row-level export)
You can select one or multiple documents to apply actions.

Working with columns

In the table view you can:
  • Reorder columns
  • Resize columns
  • Sort by any column
  • Filter by any column
  • Search by document or field value
This makes it easy to spot errors, find edge cases, and focus review where it matters.

Objects in table view

Fields of type Object (Subtable) appear as grouped items. To review them:
  • Expand the object
  • View all items in a dedicated table
This allows you to review repeated data like line items separately.

Page view

Reviewing one document at a time

The page view is designed for detailed, per-document validation. The layout is similar to the workflow creation view:
  • Document viewer on one side
  • Extracted fields on the other
The key difference:
Here, you are validating final results, not improving prompts.

Visual grounding and evidence

Each extracted value is visually linked to the document:
  • Colors and numbers connect fields to evidence
  • Hovering highlights the source in the document
This helps you understand why a value was extracted.

Confidence indicators

Each field includes a confidence score.
  • Shown as a percentage
  • Explained on hover
  • Low confidence values require human review
The goal is to reduce human effort by telling you where to look.

Validating results

You can validate:
  • Individual fields
  • Entire documents
  • Multiple documents at once
Validation options include:
  • Thumbs up - correct
  • Thumbs down - incorrect
Once validated, the document status updates and results are marked as reviewed.

Keyboard shortcuts

To speed up review:
ShortcutAction
Shift + ← / →Navigate between documents
Ctrl + EnterValidate the current document
Shift + EnterValidate and move to the next document

Correcting outputs

When a value is incorrect:
  • You can correct it manually
  • The corrected value is saved as the validated result
For Object (Subtable) fields:
  • You can validate individual items
  • Or validate the entire object
You can also:
  • Add new items
  • Filter to show only items visible on the page

What’s next?