Getting Started with MuleSoft Intelligent Document Processing

7/18/2025

Introduction

The enterprise landscape is drowning in documents—PDFs, handwritten forms, invoices, contracts, receipts, and compliance records. Yet most organizations still spend thousands of manual hours extracting data and moving it into their systems. MuleSoft's Intelligent Document Processing (IDP), which became generally available on April 29, 2024, brings a transformative capability to this space by enabling end-to-end, AI-assisted document automation directly within the MuleSoft ecosystem.

IDP allows teams to automatically ingest, interpret, and extract structured information from unstructured documents using a powerful blend of Einstein AI models and AWS Textract under the hood. While the interface abstracts the complexity for users, the backend combines high-accuracy OCR, natural-language understanding, and layout analysis to identify fields, tables, handwritten text, and contextual entities.

Once IDP extracts the data, it becomes a first-class citizen across the MuleSoft automation suite—ready to be orchestrated through Anypoint Platform APIs, transformed through Flow Orchestration, or operationalized using MuleSoft RPA bots. This drastically reduces manual workloads, accelerates business processes, and ensures greater consistency and governance across document-centric workflows.

In this guide, we'll walk you through the foundational steps for extracting data using IDP and demonstrate how to integrate the processed output with your Anypoint Platform services. Before diving into the implementation, let's first review the prerequisites required to build your IDP-enabled integration.

Step 1: Configuring User Access and Permissions

Before working with Intelligent Document Processing, you must set up the appropriate access for the user. Follow these steps:

1. Open Access Management and go to the Users list.

2. Select the user who should be allowed to work with IDP.

3. Navigate to the Permissions panel and click Add Permissions.

4. In the Document Actions category, choose all the permissions the user will need.

5. Make sure the correct Business Group is selected for the user's access.

6. Review the selections and apply the updated permissions.

After these permissions are in place, the user can access the IDP section of Anypoint Platform and begin using its capabilities.

Step 2: Creating Your First Document Action

To begin extracting structured data from documents, you need to set up a Document Action in IDP. A Document Action defines the full processing workflow—from uploading the file to running extraction, configuring fields, and applying AI prompts.

Follow these steps to create your first one:

1. Click Create New to start building a new Document Action.

2. Choose the type of Document Action you want to create.

3. Provide a name and description, then click Create.

As of the Spring 2024 release, IDP supports three document types: Generic, Invoice, and Purchase Order documents.

Uploading and Processing Documents

Once the action is created:

4. You'll be asked to upload the sample files you want IDP to analyze.

5. Upload your document and click Run to start the extraction process.

Behind the scenes, AWS Textract handles the OCR and layout analysis. The extracted values are mapped to predefined fields, which you can refine using the following options:

- Use the scan icon to highlight where each extracted value appears in the document.

- Check the confidence score to understand how reliable the extraction is.

- Mark fields as required so they must always appear; missing required fields will trigger a review.

- Show or hide fields with the eye icon to remove sensitive or empty values from the output.

Enhancing Extraction with Custom Fields and AI Prompts

You can further enhance the extraction by adding:

Custom fields, if the standard ones do not cover your data needs.

AI Prompts, which allow IDP to pull additional information based on a natural-language instruction:

- Add a prompt name and description.

- Re-run the Document Action to generate the new value.

- Manage the prompt output by viewing, hiding, editing, or deleting it.

Example:
If you want to extract a customer's PAN number, create a prompt describing the requirement and execute the action again to fetch the value.

Adding Human Review and Publishing

Before finalizing the Document Action:

- Add at least one reviewer to validate the extracted data.

- Human review ensures accuracy, builds confidence, and helps maintain data quality.

Once all configurations are complete:

- Save the Document Action.

- Publish it to Anypoint Exchange and RPA so it can be used across your integrations and automations.

Step 3: Testing Your Document Processing API

Once the Document Action is published, it becomes available in Anypoint Exchange as a REST API. Here, you'll see the auto-generated POST and GET endpoints that IDP exposes.

Testing the POST Endpoint

To verify that everything works as expected, start by testing the POST request:

1. Open the POST endpoint in Exchange and select the appropriate Server.

2. In the Body section, upload the document you want to process.

Setting Up Security Credentials

Before sending the request, you need valid security credentials:

3. Go to Access Management → Connected Apps and create a new Connected App.

4. Add the required Document Actions scopes and choose your Business Group.

5. After the app is created, copy the Client ID and Client Secret.

Then return to Exchange:

6. Enter the Client ID and Secret.

7. Expand Advanced Settings and set the credential location to Authorization Header.

8. Generate an access token and click Send.

A successful POST call returns:

- HTTP 200 OK, along with

- The Document ID and Document Name in the response body.

Testing the GET Endpoint

Next, use the Document ID or Execution ID to test the GET endpoint:

1. Open the GET method.

2. Enter the Execution ID.

3. Provide the same security credentials used during the POST request.

4. Click Send.

If everything is working correctly, you will receive:

- HTTP 200 OK, with

- A response payload containing all the extracted fields from the document.

Next Steps: Integrating IDP Across Your Automation Stack

Now that you've successfully created a document action and tested the IDP REST API endpoints, you're ready to scale your document automation initiatives. IDP integrates seamlessly with:

Anypoint Platform for building enterprise-grade integration flows that consume extracted document data.

MuleSoft RPA for creating bots that submit documents to IDP, retrieve results, and populate downstream systems automatically.

Flow Orchestration for designing visual workflows that route document data through approval chains and business processes.

The combination of intelligent extraction, API-led connectivity, and automation creates a powerful foundation for eliminating manual document handling across your organization. Whether processing invoices in finance, claims in insurance, or onboarding documents in HR, IDP transforms unstructured documents into actionable business data.

Conclusion

MuleSoft's Intelligent Document Processing represents a significant leap forward in how organizations handle document-intensive workflows. By combining AI-powered extraction with enterprise integration capabilities, IDP eliminates the manual bottlenecks that slow down business processes and introduce errors.

The steps outlined in this guide provide the foundation for implementing document automation—from configuring permissions and creating document actions to testing APIs and integrating with your broader automation ecosystem. As you expand your use of IDP, you'll discover opportunities to automate processes that previously seemed too complex or variable to handle programmatically.

The future of enterprise automation lies in making unstructured data as accessible and actionable as structured data. With MuleSoft IDP, that future is available today.