Unlocking Smarter Document Processing Through Best-of-Breed OCR and Intelligent Capture

In modern enterprise environments, document processing pipelines face increasing demands. Organisations must ingest vast volumes of documents—structured and unstructured—while maintaining accuracy, speed, and compliance. Two powerful platforms in this space are Klippa DocHorizon, known for its advanced OCR and document understanding capabilities, and Tungsten Transformation (formerly Kofax Transformation Modules), a leading intelligent document capture and classification platform.

This article explores how and why you might consider integrating Klippa DocHorizon with Tungsten Transformation, what architecture patterns to consider, and step-by-step guidance to implement a robust hybrid capture solution.

Why Integrate Klippa DocHorizon with Tungsten Transformation?

Before we go any further, let’s address the elephant in the room. You may be aware that Tungsten Transformation is now considered to be a “legacy” product. Therefore, integrating it with DocHorizon may seem to be counter-intuitive, so why even consider it?

We understand that Tungsten Transformation and Capture, although legacy products,  will continue to be supported for the next five years. Many customers will have made significant investments in their solutions and will require time and space to explore and decide the best way forward for their individual needs. Tungsten Transformation offers a fully configurable, modern HTML5-based thin client user interface which when coupled with Tungsten Capture’s wide range of document ingestion and export options is still compelling for many organisations. What it does not offer is a modern cloud-based AI engine which can natively process challenging documents with superior results. For example, look at these handwritten notes:

And look at the output from DocHorizon:

“hello this is my writing. If you can read this then fair play to you!”

“I write in frustration that I am now unable to connect with your Service Centre representatives as I did in the past. I believe that you’ve blocked my phone number and will be raising this with my MP and local councillor. I’m very angry that I’ve needed to put pen to paper and demand that my case is re-opened. Please call me or drop a live with your proposed plan of action.”

 Anyone who has tried to implement cursive handwriting extraction from documents will really appreciate the progress that has been made in the last 18 months. We were able to produce these results from these demonstration documents in less than 15 minutes! No image cleanup was necessary (the first sample looks crumpled and the lack of contrast between the text and the page colour would be challenging for many of the previous-generation OCR engines, hence the use of image cleanup tools like Kofax VRS would have been necessary.) Images processed through DocHorizon require no cleanup.

Invoice Processing

Now, let’s turn to another use case, namely processing invoices in Transformation. Transformation has many tools available for invoice data extraction, but the most used one is Specific Training, in which the system is “taught” that for a given invoice layout, “the Invoice Number is in this location, the Date in that location, etc.” This works fine until either the data moves to a different location on the page, or a user erroneously trains the system to find the same piece of information in a different location. This leads to conflicts in the training set and over time, these conflicts build up and many require many hours of resolution in order to keep the system at optimum performance.

Both platforms are strong individually, but they excel in different areas:

• Klippa DocHorizon

• Advanced, cloud-based OCR and machine learning-driven document understanding.

• Excellent for high-accuracy data extraction from semi-structured and unstructured documents (invoices, receipts, contracts and any document written in cursive handwriting).
• Scalable API-first architecture.

• Tungsten Transformation/Capture

• Enterprise capture workflows with existing integration into ERPs/line-of-business systems.

• Strong support for batch processing, zones, and verification steps.

• Mature connectors and transformation capabilities.

By integrating them, you can leverage:

• Klippa’s extraction accuracy for complex document types,
• Kofax’s workflow orchestration and system connectivity,
• A hybrid architecture that improves throughput and greatly reduces manual review.

The Integrated Workflow

At a high level, the workflow looks like this:

Key architectural points:

• REST API integration: Kofax calls Klippa REST endpoints for OCR/extraction.
• JSON document exchange: Results are returned to Kofax as structured JSON.
• Workflow branching in Kofax: Based on extraction confidence, route to verification lanes or automation.

Prerequisites

Before integrating, confirm that you have the following:

In Klippa DocHorizon

• API credentials.
• DocHorizon Flow which accepts a file from Kofax, processes it and returns the result, e.g.:

• A DocHorizon Prompt Builder configuration which contains a list of invoice fields and an AI Prompt to extract data from each one, e.g:

In Tungsten Transformation

• A functioning Transformation environment with an existing project.A script which runs in the Document_AfterExtract event and consumes the Klippa DocHorizon API endpoint to invoke the Flow.
• A script which parses the JSON returned by Klippa and populates the document fields, e.g.:

 Benefits Realised

For existing Kofax customers, marrying Klippa’s powerful extraction engine with Kofax’s workflow orchestration, organisations can attain:

• Reduced manual data entry,
• Higher extraction accuracy,
• Faster processing cycles,
• Lower operational costs,
• Seamless integration into existing systems.

Final Thoughts

If you are an existing Kofax/Tungsten Transformation / Capture customer, now is an optimal moment to consider modernising your capture solution. Integrating Klippa DocHorizon with Tungsten Transformation is one way of creating a forward-looking capture ecosystem and providing a springboard into the future. It provides you with a halfway house that blends best-in-class OCR with enterprise-grade workflow control. It will extend the life of your capture system, whether you are digitising invoices, forms, or correspondence, while you consider what your final move will be. Stay tuned for more thoughts on this subject!

If you would like a demonstration or pricing, please contact us.