File split is a type of AI Processing. It automatically splits batch PDF files – scanned bundles containing several documents – into individual AODocs documents, with no separator pages and no manual cutting. 

This article explains how to configure AI-powered File Split in your AODocs libraries to automate document splitting and reduce manual workload.

Note: AODocs File Split is available only in Secured Folders and Document Management libraries. It isn't available in Team Folders.

Automatically generated table of contents

What is File Split?

Scanned bundles and batch uploads often contain several documents in a single PDF. Splitting them manually is time-consuming, tedious, and error-prone.

With File Split, the AI analyzes each batch PDF, detects the boundaries between documents, classifies each part against the document types you configure, and creates one AODocs document per part in its target document class – with a confidence score for each proposed section.

Batches where the AI is highly confident are split automatically. The others are routed to a review step, where your reviewers validate and adjust the proposed split before it's executed.

Prerequisites

Before you begin, you must:

• Make sure that the File Split capability is set up on your domain and activated in your library.

For information on setting up and activating AI for AODocs, contact the AODocs Support team by email at support@aodocs.com or open a ticket.

During this process, the AODocs Support team will allowlist a service account on your domain. 

• Activate the new library administration in your library – learn more: 
  • Activate or deactivate early-access features in libraries (for library administrators)
  • Activate or deactivate early-access features on your AODocs tenant (for super administrators)
     
• Add the allowlisted service account as a library administrator in your library. To do this, do one of the following:
  • open the library's security settings and add the service account provided by AODocs Support as a library administrator – learn how to do this in Secured Folders and Document Management libraries
  • click Continue in the dialog that opens the first time you use AI Processing

Create AI Processing configurations

You can create as many File Split configurations as you like in your library. Each configuration has its own intake document class, workflow, and views – so you can create one configuration per kind of batch you receive. For example, one configuration for delivery note bundles and another for scanned contract files.

1. Open the library administration.

2. Click the Create Configuration button and select File Split.

3. Enter a name for your AI processing configuration. In our example: "Delivery notes batch"

4. Click Create to begin configuring the AI processing.

In the background AODocs starts setting up everything you need to start testing the feature:

• A fully functioning workflow specifically for managing split documents, with the following states:

• A document class to ingest your batch documents. It has the same name as the file split configuration you entered earlier. In our example: "Delivery notes batch".
• A dedicated role - reviewers - allowed to use the Human In the Loop Interface and review the splits suggested by AODocs AI.

Note: This role is created without any members. Don't forget to add users to the role.

• A set of views to list the documents to be reviewed, the processed documents…

Configure the AI processing

Now that all the configurations you need on the library have been created, you can start configuring the AI File Split processing.

1. Click the Add section button.

A panel opens where you can define the different sections. Each section corresponds to a document type. For each section, enter a precise name – it helps the AI recognize that part reliably – and define where identified documents are dispatched: the target library and document class.

In our example, we have: 

• delivery note (note from the company shipping the goods)
• receipt notes (note from the company receiving the goods)
• notes from the transporter (CMR)

screen: The Documents to process panel with the three example sections defined, each dispatched to its target document class.

screen: The section configuration panel with the section name, the optional instructions, and the 'When identified, dispatch to' library and document class selectors.

Important: 
– Section names must be unique. However, you can define the same document class to as many sections as you like.
– The document class can be defined in a different library from the one where the File Split is configured.

2. Optionally, click Add instructions. 

Instructions can be used to provide additional context to the LLM to help it differentiate between the sections. 

This is not mandatory. We recommend testing first without instructions.

3. Click Save.

4. To activate the AI processing click then on Set as active.

Additional settings

You can access some additional settings by clicking on the slider icon on the top right hand corner.

screen: The settings panel with the global instructions, the pages batch size and overlap size, and the Force validation on every section option.

Add instructions to make processing more relevant

By activating this setting you can add another instruction field this time covering all the sections. This setting can be useful to help the user understand the kind of document ingested and increase the overall confidence score generated.

Define max number of processed pages 

By activating this setting you can change batch size (50 pages by default) and Overlap size (10 pages by default).

The File Split processing can manage files up to 4000 pages. But the actual context window of the LLMs does not allow to process it in one batch.

The tool therefore processes big files in batches with a certain overlap to ensure we can spot changes.

Adjust these settings if, for example, your largest documents are around 60 pages, or if the documents you split are only 1–2 pages long. Smaller batches improve accuracy but make processing longer.

Force validation on every section

You can activate this setting if you want your team to first validate all the splits generated by the AI. We recommend starting with this setting activated.

Test your configuration

Use the Preview button to test your configuration before processing real batches. You can select a document from the library or upload a test PDF.

The preview runs the real AI processing and displays the proposed sections with their confidence scores – in read-only mode. Nothing is modified: no documents are created, the workflow states don't change, and uploaded test files aren't added to the library.

screen: The preview dialog showing a test document and the sections proposed by the AI with their confidence scores.

What happens when a batch file arrives

When a document with a batch file is created in the intake document class:

• The AI analyzes the file and proposes a split with a confidence score per section.
• If every section has a High confidence score – and you didn't force validation – the file is split automatically and the documents are created in their target classes.
• Otherwise, the batch moves to the Action Required state and appears in the Pending Review view, where reviewers validate and adjust the split.
• Processed batches appear in the Completed Batches view. Failures appear in the System Failures view, and an error notification email is sent – the workflow's retry actions relaunch the failed step.

Important: The batch document must have exactly one attached file, and it must be a PDF. Otherwise, the processing isn't triggered.

Related articles

• Review and validate AI file splits
• AI for AODocs: Set up AODocs Intelligent Document Processing in your library