ABBYY purpose built AI center

ABBYY previewed a suite of new innovative capabilities for purpose-built AI that empower businesses to simplify success and accelerate ROI in their automation initiatives. These include Phoenix, a multi-modal approach to zero-shot learning leveraging small language models, purpose-built for document tasks, a Secure Large Language Model (LLM) Gateway, and enterprise-ready tools and applications that enable global enterprises to gain more value and insight into business-critical document processes. Announced at ABBYY Ascend 2024, the latest AI innovations are centralized in the new ABBYY Purpose-Built AI Center to help innovation leads, and developers discover the array of tools available from ABBYY to accelerate application building.

ABBYY purpose built AI center

The Purpose-Built AI Center serves as a centralized knowledge base for businesses to access information about ABBYY's suite of AI tools to develop next-generation applications. Here, enterprises and developers will be able to learn about the range of capabilities that purpose-built AI from ABBYY offers, including how small language models are leveraged to deliver capabilities typically seen only by using LLMs to deliver a faster time to value. By simplifying and accelerating the onboarding of new document types within enterprise business processes, businesses have more agility to adapt to evolving needs. This comes in addition to the already available 80+ document models available out of the box.

Using ABBYY intelligent document processing (IDP) as a secure LLM gateway

By using ABBYY intelligent document processing (IDP) as a secure LLM gateway, businesses can command an LLM to extract data while simultaneously validating that the data is present on the document at hand. This empowers organizations to harness the power of general-purpose LLMs while significantly limiting hallucinations and increasing the reliability and trust of the output.

ABBYY also highlighted a collection of capabilities designed to improve efficient operations for large enterprises:

• E-invoice processing: A streamlined, compliant processing of all invoice formats, ensuring adherence to diverse national standards.
• Quality analytics: Tools for trend analysis of straight-through processing (STP) rates, continuously improving data extraction quality and performance.
• Big data analysis: Enhanced scalability for process analysis, capable of handling complex, event-driven data with billions of data points.
   

"With the launch of our Purpose-Built AI Center, we are making it easier for our customers to find the right solution for their business needs, simplifying their success in automation. ABBYY’s innovations truly provide a response to the main concerns businesses have in the market today when leveraging AI and doing so at scale within an enterprise context," said Maxime Vermeir, Senior Director of AI Strategy at ABBYY.

ABBYY also announced its latest optical character recognition (OCR) performance improvements and tools equipping developers to more efficiently create cutting-edge applications. Highlights include:

• Enhanced OCR capabilities with better memory consumption, improving document conversion quality and document structure detection.
• New handwriting recognition (ICR) for multiple languages, document comparison modules, and updated processing profiles for data extraction, natural language processing (NLP), and archiving.
• Simplified API support, as well as new and updated support for popular development languages such as .Net, Python, and more.