Supercharging Intelligent Document Processing with Generative AI

Gone are the days when document processing was synonymous with endless paperwork and manual labor. Now, with AI, documents are gateways to automation. The global Intelligent Document Processing (IDP) market, now at an impressive USD 2-3 Bn, is projected to reach USD 10-13 Bn by 2027. Driving this growth are the increasing volume of documents, a need for time optimization, and a focus on employee and customer experience. While not entirely new, IDP technology has been revolutionary for sectors like Insurance and Banking. Its evolution has been rapid, transitioning from manual processes to Intelligent Automation.

How is Generative AI Reshaping IDP?

The advent of Generative AI in the IDP realm has been nothing short of transformative. This technology, too, boasts a rich evolution, starting with the early probabilistic models of the 2000s and scaling to groundbreaking innovations like Variational Autoencoders and Generative Adversarial Networks of today. The unveiling of Transformer Architectures and Large Language Models (LLMs) such as OpenAI’s GPT series, heralded a fresh chapter for Generative AI. This shift has enabled researchers to fine-tune models like GPT for specific tasks, including intricate document analysis.

Delving a bit deeper, the influence of Generative AI on IDP becomes even more evident. Advanced neural networks and generative models have the power to elevate document pre-processing to unmatched levels. Tasks like noise removal, enhancing resolution, and boosting text legibility have been transformed. Document classification, too, has achieved new heights, courtesy of synthetic labeled data generated on demand. In the realm of data extraction, natural language prompts steer LLMs to pinpoint and extract pertinent information with finesse.

What’s more, synthetic datasets facilitate rigorous validation with diminished manual oversight. One of the most tantalizing prospects of Generative AI is its potential to auto-populate missing data, promising to revolutionize data integration. And when it comes to analytics, these advanced models can distill and visualize document insights through simple text prompts, streamlining the process.

Casting a Spotlight on Kofax’s Journey

One of the trailblazers in document processing is Kofax, which has been in the business for over 35 years. They have more than 25,000 customers spread across the globe, majorly focusing on Banking, Financial Services, Insurance, Public Sector, Supply Chain, and Healthcare industries. Kofax has consistently showcased its commitment to innovation, particularly in the realms of Artificial Intelligence (AI)/Machine Learning (ML) and Robotic Process Automation (RPA). Their illustrious history, marked by strategic acquisitions, has solidified their position at the forefront of both the IDP and the broader Intelligent Automation segments.

While the Generative AI wave has been ongoing since November 2022, we saw all the Automation players lining up to ride this wave and grab customer mindshare with newer Generative AI product offerings. However, Kofax has been very cautious yet deliberate about their foray into Generative AI. Rather than jumping onto the Generative AI bandwagon to ride the hype, they went back to their vast customer base to seek inputs on how best to go about this new technology, while addressing compliance requirements and other concerns. Kofax spent a considerable amount of time consulting with their customers before coming to market and announcing their latest innovations last week:

• Kofax has introduced an OpenAI Connector, which is available in its marketplace today. This connector, rather than being a mere feature, is a transformative tool. It facilitates integration with both Azure OpenAI and OpenAI, granting organizations the capability to harness Generative AI for tasks like content summarization, translation, and content generation, all within Kofax TotalAgility (KTA) processes.
• Looking ahead to the future needs of developers and businesses, we learned that the upcoming release of Kofax’s TotalAgility platform is set to feature an in-built activity that represents an evolution of the existing OpenAI Connector. This activity will be natively integrated within the platform, seamlessly allowing developers to embed Generative AI capabilities directly into their processes. Moreover, it promises flexibility, supporting integration not only with OpenAI and Azure OpenAI, but also with any LLM of the customer’s preference.
• A ‘Generative AI CoPilot’ is another feather in Kofax’s cap. This tool is set to redefine developer productivity. It leverages Generative AI to simplify and automate the creation of workflows, data models, forms and more within the KTA platform, all based on natural language inputs. It’s a leap beyond Low Code –it’s the evolution of Automation.
• Zero training extraction, powered by Generative AI, is another flagship offering in the pipeline. This feature will expedite and simplify the design of document extraction models. It promises to reduce the need for ongoing training while delivering enhanced extraction accuracy. The days of spending significant amounts of time up front on training and model design will soon be a thing of the past. Integrating Generative AI will make this process so much faster and easier for organizations.
• Additionally, Kofax’s forward-thinking approach is evident in their focus on blending the use of LLMs with Retrieval Augmented Generation (RAG). In collaboration with a strategic partner, Kofax is in the process of developing a connector that combines the power of Generative AI with RAG. This strategy aims to enable organizations to produce personalized insights based on their proprietary data (via RAG) and augment it with the broader contextual understanding that public LLMs offer.

Besides these newer features and offerings, Chris Huff, Kofax’s Chief Growth Officer, also shared with us that they have recently onboarded a new SVP of Global Partnerships, Eric Magnuson, and his charter would be to expand ecosystem activity that relates to Generative AI, in building deeper inroads with strategic partners including but not limited to those of Google and AWS. Kofax will work on what Chris referred to as the ‘Bring Your Own License’ or BYOL model where the customers can choose the LLM of their liking, and Kofax will integrate it within their existing KTA workflows seamlessly. This will surely be a gamechanger in the industry!

“Customers are at the core of everything we do at Kofax. For this reason, we kickstarted our Generative AI journey by engaging our customers and building a Customer Advisory Board to assist in designing our Generative AI strategy. From the beginning, our strategy focused on developing powerful AI-led Automation to solve our customers’ most challenging digital business needs. We are proud to have recently unveiled our Generative AI approach, which includes partnerships with hyperscalers, LLM connectors, and a soon-to-be released Generative AI-infused TotalAgility platform, which we believe will once again disrupt the IDP market!”

Chris Huff, Chief Growth Officer, Kofax

What are other IDP players doing?

Generative AI’s emergence as the next frontier has seen leading IDP platforms rush to integrate its capabilities into their existing offerings. These platforms are capitalizing on synthetic data generation, natural language processing, and innovative problem-solving, setting new standards in the IDP sector.

What does the future look like for IDP?

Generative AI’s ascendancy marks a significant shift in the IDP realm. As it matures, its integration into document processing workflows is only going to intensify. The early benefits are evident – enhanced processing, classification, and extraction. Generative AI’s contextual understanding and natural communication skills promise to revolutionize document intelligence extraction. While there is room for evolution, the future of document processing, with generative models at the helm, looks brighter than ever.