How Ocrolus works

See how Ocrolus optimizes your workflows for processing financial documents

Welcome to the guide section of Ocrolus, where we'll explore how this platform can help you optimize your workflows for processing financial documents. This guide will provide a detailed understanding of how Ocrolus works and how it can benefit your organization.

Ocrolus provides an API that seamlessly classifies documents, captures key data fields, detects fraud, and analyzes cash flows and income. In the following sections, we'll explain each step of the process in detail:


Using advanced machine learning algorithms, Ocrolus can quickly recognize and categorize your documents. In case of any uncertainties, our system automatically routes the document to Ocrolus verifiers for human review, ensuring the highest level of accuracy. To learn more about Classify, see Classify product guide and API reference documentation.


Ocrolus extracts data from your document and returns the data to you in a structured format, eliminating the need for manual entry and validation. This significantly reduces labor hours and accelerates the processing time from end to end. To learn more about Capture, see Capture product guide and API reference documentation.


Ocrolus checks qualifying documents for signs of suspicious activity, such as file tampering, incomplete data sets, missing pages, invalid dates and amounts, abnormal fonts, and irregular formats. Our fraud detection algorithms work around the clock to keep your financial documents secure. To learn more about Detect, see Detect product guide and API reference documentation.


Ocrolus helps you gain deep insights into borrower cash flows and income with clean, normalized data. It allows you to make informed decisions for your business with ease by analyzing financial data. To learn more about Analyze, see Analyze product guide and API reference documentation.

Processing Types

The processing types are directly linked to the book_class parameter in the Book creation and Book update API requests.


Instant Books undergo processing solely through automated means without any human intervention. This method ensures swift processing speeds and shorter turnaround times.



  • Not all document types are supported for Instant processing. To know more about the supported documents, see the list of supported documents.
  • The subtotal fields and values outlined here will be null for pay stubs that are processed through Instant. This is because the line items for earnings and deductions on Instant pay stubs are not captured.

For the supported documents, fully classified, captured, detected, and analyzed data is available to provide comprehensive results. Whereas, for the unsupported documents, their status will be indicated accordingly, allowing you to switch the processing type from Instant to Complete for entire Book processing.

To create an Instant Book, set the book_class parameter to INSTANT while creating a book using Create Book API.
To change the processing class from Instant to Complete, update the book_class parameter to COMPLETE while updating the Book using Update Book API.

By default, Books will be created with the COMPLETE processing type assigned as their book_class. Analytics on Instant Books can only be accessed through the updated endpoints. For more detailed information, refer to the following links:


Complete Books are processed using the standard approach of initial machine automation followed by human-in-the-loop validation and verification when required. While this method may be slower compared to Instant processing, it ensures a higher level of accuracy, with a guarantee of 99% accuracy. This method also helps you process a broader range of document types.

To create a Book as Complete, you can either set the book_class parameter to COMPLETE while creating a Book using Create a Book API or omit the book_class parameter entirely. If you omit the book_class parameter, the book will undergo processing using the standard human-in-the-loop approach as the default method.

Recommended Workflow

If you wish to utilize both Instant and Complete processing, we recommended using the following workflow:

  1. Create the Book as Instant initially and upload the documents.
  2. Check the Book results using the Book Status endpoint to identify which documents were successfully processed through Instant. You can configure webhook with verification_complete event to know the status of the Book processing. To know more about webhook configuration, see Configure webhook endpoint and Organization level webhook guide.
  3. Review the metrics calculated from the uploaded documents using the following endpoints:
  4. If enough documents for gathering information and calculating metrics to make a decision on the application are successfully processed through Instant, you can proceed with the current processing. However, if an insufficient number of documents are processed successfully through Instant or the metrics fall below the desired threshold, upgrade the Book to Complete using the Update Book API. Then, reassess the metrics based on the new processing method.