The Company

ABBYY was founded in 1989, making it a forerunner of the document capture industry. For more than 30 years, ABBYY was (alongside Kofax) the best-known name in the document capture industry. As of today, it claims over 100 million users worldwide.

ABBYY was founded as one of the first optical character recognition (OCR) software providers and has grown its product line as well as its international footprint to include offices in the US, Europe, and Asia. In 2019, ABBYY expanded its software beyond capture into intelligent document processing (IDP), process mining (ABBYY Timeline), and intelligent process automation.

CEO Ulf Persson leads the firm, and Marlin Equity Partners has been the majority owner since May 2021. We estimate that ABBYY’s revenues are approximately $200 million and that it employs around 1,300 people worldwide, with headquarters in Charlotte, NC.

The Technology

ABBYY has long been nearly synonymous with its document capture products FlexiCapture and FineReader, which have become industry standard products worldwide. However, this report focuses on its intelligent document processing (IDP) product, Vantage.

ABBYY Vantage was launched in 2019 as the company’s foray into IDP, and it came to fruition with 2021’s launch of Vantage 2, which provides a no-code platform to access the underlying machine learning (ML) functionality. The launch of Vantage 2 coincided with the launch of its associated online Marketplace.

Vantage goes beyond capture and aims not just to extract data from documents but also to further process that data through the use of machine learning. The Vantage platform comprises many pre-trained ‘’AI Skills” accessible to users through a low-code/no-code environment. Users also use Vantage Skill Designer, leveraging the platform to build their own AI skills modules.

More commonly, users will access pre-configured modules and further train them on their data through the associated ABBYY Marketplace, which offers around 150 document-related models delivered as Vantage Skills (see Figure 1). These pre-trained skills range from specific document types such as 1040 Schedule C tax forms or a Medical Examiner’s Certificate to others that can automate data classification and capture from bills of lading, invoices, receipts, identity documents, and more. The range and depth of skills in the Marketplace are impressive and have snowballed since its 2021 launch.

To dive deeper, it’s worth taking a detour to underreported and, frankly, underappreciated elements of ABBYY’s work. NeoML, its open-source machine learning foray, was quietly launched in 2020. Available through GitHub, NeoML is a framework for developers to access traditional ML and more complex neural network (deep learning) based algorithms. Another way to think of NeoML is as a readily accessible library of components to design, build, and deploy ML models.

NeoML is far from the only open-source framework or library around, but it is highly specialized, focusing only on image and text processing. The elements included within NeoML were designed by ABBYY and claim to offer significant advantages over the other options (most commonly TensorFlow) in terms of improved speed, lower memory and processing demands, on-device processing, etc. All in all, NeoML is impressive, though clearly, this C++ environment is not for amateurs or non-developers. But it does give insight into the remarkable work that lies beneath the surface of Vantage but does not require a Vantage user to be an ML expert.

In summary, ABBYY is treading an impressive line between low-code/no-code and highly complex and advanced algorithms. With Vantage, on the surface there is an easy-to-use no-code platform and Marketplace that gives organizations access to a wide range of specialized pre-trained modules that they can quickly deploy. But underneath, there is another world of advanced AI, much of which has been shared through open source to developers. That’s a significant difference from many of its competitors, which go to some lengths to mask how their IDP systems work from buyers and analysts alike.
Finally, it’s important to note that while Vantage is and will be a core component of many IDP projects, it does not provide enterprise automation (EA) capabilities. Instead, it connects to commonly used EA products such as Pega, UiPath, Blue Prism, Automation Anywhere, and many other automation systems. Rather than a limitation for Vantage, this neutrality regarding the selection of automation systems (and, for that matter, data sources) is arguably a positive for those wanting to add advanced capture and document processing to existing and new best-of-breed environments.