Split single PDF with Multiple invoices(different formats)

Question

Hi,

We have a requirement to split a single PDF file into multiple invoices. The PDF contains invoices in different formats and varying page ranges, without following a standard template.

Example : single PDF --> split into 3 PDF's each as separate Invoice in different formates, etc.

Could you please suggest any possible approach to achieve this?

Regards, Satya

Answer 1

Hi Kapaganti, Satya,

Based on your scenario, using Azure Document Intelligence’s Custom Classification Model is a valid and promising approach for handling the splitting of multi-invoice PDFs, especially when invoice formats vary.

The Custom Classification Model is designed to classify documents based on their content and layout, and it can be trained to identify the start of different invoice types or boundaries by labeling representative samples. Once trained, the model can process each page or document section and assign a classification label, effectively indicating invoice segments. However, while this enables classification of content, the platform currently does not support automatic PDF splitting based on classification results as a native low-code feature. You would still need to implement a small amount of logic (e.g., using Azure Functions or Logic Apps) to parse the classification output and then split the original PDF accordingly using tools like Azure Form Recognizer Studio, REST API, or PDF libraries in Power Automate. So, while the classification model reduces the complexity of custom rule development, a hybrid solution that combines Azure’s intelligent classification with lightweight automation is currently the most practical and scalable path. This approach remains largely low-code and supports easier future maintenance through model retraining, without requiring heavy custom development.

Answer 2

Hi Pavankumar Purilla,

Thank you for your detailed response and the suggested approach.

As you rightly pointed out, the invoices in our PDF files do not follow a fixed template or consistent page count, which makes rule-based splitting challenging. The method you suggested using text extraction and layout analysis—especially with Azure Document Intelligence—is indeed in line with what we were considering.

We wanted to check specifically whether this use case could be addressed using Azure Document Intelligence → Custom Model → Custom Classification Model. That way, we could train the model to detect and classify invoice boundaries even when formats vary.

Our preference is to explore a standard or low-code approach rather than a fully custom-coded implementation, if possible. This would help us streamline future maintenance and scalability. Please advise if such classification-based splitting is achievable using Azure’s prebuilt/custom features—or if a hybrid, service-composed model is the only way forward.

Looking forward to your input.

Regards,

Satya