ABBYY expands document AI patent portfolio as GenAI grows
ABBYY has secured 22 new patents covering techniques used in document processing, citing growing interest in specialised approaches to handling business data as generative AI adoption expands.
The additions take ABBYY's portfolio to more than 400 patents and patent applications worldwide. The company develops software for document process automation and related analytics, with offices in Singapore and Australia alongside its headquarters in Austin, Texas.
The newly issued patents cover language detection, user interface design, image identification, information extraction, and document decoding. ABBYY said the work behind them took place over five years, and the patents were issued over the past two years.
Document Focus
Several patents relate to extracting and structuring information from document images, a common requirement in sectors that handle large volumes of forms, correspondence, invoices, claims documentation, and identity material.
ABBYY highlighted patents including "Extracting Multiple Documents from a Single Image" and "Detecting Fields in Document Images". These methods sit within a broader field that combines optical character recognition with classification and data capture.
ABBYY also pointed to patents covering associations between data points in a document, including "Identification of Key-Value Associations in Documents Using Neural Networks". Another, "Continuous Learning for Document Processing and Analysis", relates to systems that adapt as document formats and inputs change.
The batch also includes user interface work. ABBYY cited "Display Panel or Portion Thereof with a Graphical User Interface", reflecting a focus on how document automation tools present information for review and exception handling.
Image Decoding
Two of the patents focus on processing images and barcodes in less controlled conditions: "Decoding of Two-Dimensional Barcodes Under Unfavorable Conditions" and "Assessment of Image Quality for Optical Character Recognition Using Machine Learning". These issues often arise when organisations capture documents on mobile devices or ingest scans that vary in resolution, lighting, and alignment.
Document AI suppliers have increasingly positioned image quality assessment and robust barcode decoding as part of front-end capture. Poor input quality can increase manual review and raise the risk of extracting incorrect data.
GenAI Risk
ABBYY framed the new patents in the context of enterprise use of generative AI, where businesses weigh potential productivity gains against accuracy concerns, auditability, and regulatory expectations. It described its work as "purpose-built AI" for documents and business processes, positioning the approach as a way to deliver accurate outputs aligned with compliance standards.
In many organisations, document workflows remain a key source of operational and compliance risk. Errors in extracted fields can affect downstream systems in finance, procurement, customer onboarding, and claims processing. Regulated industries, in particular, face pressure to show how automated decisions and data capture steps were produced.
ABBYY said its customer base includes more than 10,000 enterprises and cited 35 years in document and process automation software.
Software-related patent grants are often closely scrutinised. ABBYY opened its announcement by citing technology patent acceptance rates of around 55%, without providing a source.
Ulf Persson, ABBYY's chief executive officer, linked the patent awards to the company's product direction.
"ABBYY's latest patents are a testament to our ability to drive the next level of innovation. By combining cutting-edge AI with a deep understanding of intelligent document processing and business needs, we are shaping the future of intelligent automation and delivering solutions that empower businesses to thrive in an ever-evolving landscape," said Ulf Persson, CEO, ABBYY.
ABBYY plans to showcase its products at ABBYY Ascend events in Nashville and Brussels this spring.