IDP by Workato

Intelligent Document Processing (IDP) by Workato extracts data from images and PDFs, converts unstructured content into structured output, and integrates it into business systems. IDP reduces manual data entry and accelerates workflows such as invoice reconciliation, claims management, identity verification, and document processing.

FEATURE AVAILABILITY

IDP by Workato is available for direct customers on specific pricing plans across all Workato data centers. It is available to Embedded partners in the admin workspace, but is not available in Embedded customer workspaces. You can opt-in to this feature by agreeing to our AI feature addendum (opens new window). Contact your Customer Success representative to learn more.

IDP enables you to process invoices, reconcile financial records, extract claim details, verify identities, and parse structured or semi-structured documents such as resumes and receipts.

Automated document processing

Manual document handling can delay operations, cause inaccuracies, and raise costs. For example, Accounts Payable (AP) teams manually extract and verify invoice details, which can lead to errors and approval delays.

IDP by Workato automates data extraction and reduces processing time from 8-10 minutes per document to 20-30 seconds. Extracted data maps directly into ERP, CRM, and financial systems, removes manual input, and ensures high accuracy. Confidence scores and automated validation trigger human review when necessary, prevent errors, and ensure compliance.

Supported input formats

IDP by Workato supports multiple document formats and languages to support your requirements:

• File types: IDP accepts files in PDF, PNG, JPG, WebP, and GIF formats. Documents must be clear, legible, and contain less than 15 pages.
• Languages: IDP achieves high extraction accuracy when it processes invoices and receipts in English, Chinese, Japanese, and Korean, based on internal evaluations.

When to use IDP by Workato

You can use IDP to process structured documents such as invoices, receipts, and purchase orders with up to 15 line items.

You can use free-form schemas to extract and map data into ERPs, CRMs, and financial systems for unstructured documents.

IDP is ideal for businesses that process high document volumes, require structured data extraction, or need to reduce manual data entry errors.

Common use cases

You can use IDP by Workato to automate document processes across multiple business functions:

• Order to cash

• Your Sales teams can process purchase orders manually, extract details, and verify information before entering the data into NetSuite. IDP extracts and maps purchase order data automatically to reduce processing time.

• Accounts payable

• Your accounts payable team can extract invoice details manually and enter the information into Coupa. IDP automates invoice data extraction, ensures accuracy, and removes manual input.

• Procure to pay

• Your procurement team can cross-reference invoices and delivery orders with system records. IDP extracts data and identifies discrepancies for review.

• Expense management

• Your employees enter expenses manually. IDP extracts receipt and invoice data to simplify expense reporting and reimbursement.

Connection setup

This connector doesn't require connection setup.

Confidence scores and validation

IDP by Workato assigns confidence scores to extracted values to measure workflow effectiveness and identify when reviews are necessary.

A confidence score is a number between 0 and 1 that represents the likelihood that extracted output matches the expected data. Higher scores indicate greater accuracy.

Confidence thresholds vary across use cases. Test your use case, document type, and language to establish a reliable threshold and validate extraction performance. You can assign a team member to review extracted values with confidence scores of less than or equal to (≤) 0.95 to improve accuracy.

Limitations

Consider the following limitations when you evaluate IDP for your use case:

• Document size limit: IDP processes documents up to 15 pages. Larger documents require splitting with the pypdf library on the Python connector, followed by manual reconciliation.

• JSON schema constraints: IDP supports up to 100 fields with a maximum of 5 nested levels. The total schema length must not exceed 15,000 characters.

Considerations

IDP may not accurately extract small text or non-Latin characters, which can cause extraction discrepancies. The following issues may occur:

• Inaccurate values
• High confidence scores for incorrect values
• Low confidence scores for accurate values

You can help prevent these issues by setting up verification and validation checks in workflows to improve accuracy. You can also flag values for manual review if the confidence score is ≤ 0.95, or extracted fields don't match expected values.

Clear, legible documents and well-structured JSON schemas often lead to more accurate results. Complex tables and documents with long line items may reduce extraction accuracy, especially when fields span multiple rows or columns.