Salesforce Docs Portal

Define What Data to Extract from Your Documents

Document Processing Configurations define the structure and rules for extracting data from your documents. By creating these configurations, you define what information to look for and how to organize the extracted data. This step is essential before you can process documents in your flows or set up review workflows.

Required Editions

Available in: Lightning Experience
View supported editions.
This feature requires the MuleSoft for Flow: IDP add-on. Professional Edition requires the API access add-on. To purchase, contact your Salesforce account executive.
Document processing features require Einstein generative AI turned on in Setup, and Data 360 provisioned and enabled for your org.
MuleSoft for Flow: IDP features used with Agentforce require the Foundations or Agentforce 1 edition. To purchase these editions, contact your Salesforce account executive.
User Permissions Needed
To create, read, edit, and delete document processing configurations: Manage Document Processing Configurations

Before you define what data to extract from your documents, ensure you have:

  • Sample documents of the type you want to process (PDFs, images, or scanned documents)
  • A list of the specific data fields you want to extract from your documents
  • Knowledge of the data types for each field (Text, Number, Currency, or Date)
  • Understanding of the document structure and layout you want to process

Create Document Processing Configurations in the Automation App to define the instructions and output structure for each type of document to process. You can upload documents and test the extraction process while creating the Document Processing Configuration.

If you plan to use the Only when needed option in Human Review Settings, a review is created only when one or more extracted fields or table rows fall below the confidence threshold. Use the confidence scores in the test results to judge which fields are likely to need review. See Flow Core Action: Extract Data from Document.

  1. In the Automation app, select the Document Processingtab.
  2. Click New Document Processing Configuration.
  3. Click Upload Files and select an example document to use as a basis for creating a Document Processing configuration.
    Click Done when the upload finishes.
  4. Click Using Auto-Extraction to automatically add fields and tables based on the uploaded document.
    This is the recommended approach as it scans the sample document and automatically identifies and extract all found fields.
  5. Optional: If needed, manually add additional fields and tables to the configuration.
    1. In the Outputs panel, click Fields and click Add Field.
    2. Provide Name, Field Type (string, number, boolean), and optionally Prompt Instructions with additional details for Einstein to consider during the extraction process.
    3. Click Add.
    4. Continue adding fields as necessary.
    5. To add tables, in the Outputs panel, select the Tables tab.
    6. Click Add Table.
    7. Provide Name, optionally Prompt Instructions, and click Add.
    8. Click the table name and select the Columns tab.
    9. Click Add Column.
    10. Provide Name, Field Type, optionally Prompt Instructions, and click Add.
    You can add fields and tables manually after auto-extraction or choose to create the configuration entirely manually by clicking Create Manually instead of Using Auto-Extraction.
  6. Click Test to analyze your uploaded document and see the extracted table data.

    The test results show extracted values for each field and table, along with confidence scores indicating the accuracy of the extraction.

    Reviewers see the same confidence percentage for each field in the Document Review window when a flow requires human review.

  7. After you finish creating the configuration, provide a name for your Document Processing Configuration and click Save.

Your Document Processing Configuration is now saved and ready to use. The configuration defines all the fields and data types to extract from your documents. You can verify the configuration is complete by checking that:

  • All required fields are defined with appropriate data types
  • The test analysis shows accurate extraction results
  • Confidence scores are acceptable for your use case
Example
Example

Here's an example of field configurations for processing invoices:

Field Name API Name Field Type Prompt Instruction
Company Name CompanyName Text Extract the vendor or company name from the invoice header
Order Number OrderNumber Text Find the invoice or order number, usually near the top of the document
Purchaser Name PurchaserName Text Extract the customer or purchaser name from the billing section
Total TotalAmount Currency Find the total amount due, usually at the bottom of the invoice
Tax TaxAmount Currency Extract the tax amount if listed separately

This example shows how to structure field definitions for common invoice data. The prompt instructions help Einstein understand where to look for each piece of information in the document.

After you define what data to extract from your documents, add the Extract Data from Document action to your flow and set up routing to human review with Human Review Settings. See Set Up Document Extraction and Routing to Human Review.