ABBYY AI Summit 2025: Purpose-built AI for intelligent document processing
The Computer Weekly Developer Network attended the Abbyy AI Summit this week to get the low down on the digital workflow tools for developers in the document management space and beyond.
Held at the Stephen F. Austin Royal Sonesta Hotel in Austin, Texas, this event brought together automation strategists and AI practitioners to explore practical applications of AI in document processing and business automation.
A welcome session from Bruce Orcutt CMO at Abbyy kicked off the event with some notes relating to the fact that the company has not hosted a user engagement event like this for a while (there has been a pandemic, remember?) and that this event – with a hashtag of #ABBYYAISummit – is now focussed on what Orcutt calls “purpose-built” AI.
This is a term that means the company is not trying to be generative, but to focus on document management to inject valuable information into business processes. The company leans heavily on “process intelligence” which is an approach to process optimisation, workflow improvement, analytics insights, operational efficiency and productivity gains.
Bemoaning the fact that not enough AI hype today (across the tech industry and the media landscape) is focused on outcomes, Orcutt says that Abbyy is now all about outcomes and integrating with operational systems in enterprises working across all verticals.
Talking of verticals, Orcutt refers to Abbyy customers, particularly on the banking side, that are using AI to make decisions that become particularly pertinent when we’re talking about money.
IDP – Intelligent Document Processing
The market for document intelligence (let’s call it IDP, intelligent document processing) has moved from ECM to BPM to RPA… and now to a third level of disruption through the arrival of AI. This shift has changed who the buyers are for a company like Abbyy, especially in light of the now more focused involvement of AI engineers in the technology decision-making process.
Banking and financial services, healthcare, software & IT services, insurance and retail and the key markets for Abbyy in the IDP space. As the point of AI coming into its platform offering, Orcutt insists how stringent the company’s compliance rules are over AI governance to ensure models being used are free from bias. The company has a formalised “Trust Centre” which it uses to bring together its knowledge base in this area.
“The AI landscape is shifting rapidly – from prompt engineering to adaptive, generative systems that can reconfigure themselves based on context and goals. Generative algorithms are changing what it means to build and deploy intelligent systems.
Humans & Knowledge Transfer
The morning continued with a session entitled ‘The Generative Frontier: Self-Adaptive AI and the Next Wave of Strategic Innovation’ presented by Jepson Taylor in his role as CEO and founder at VEOX, a company that provides precise data acquisition, high-quality imaging and efficient workflows.
“Humans are the most adept at processing intelligence because we have the ability to use intelligence transfer i.e. we can tell somebody about something and they can understand it without having to experience the action or fact itself,” said Jepson, explaining where we ourselves stand on the spectrum of learning and intelligence management. “However smart you are, you are probably forgetting that your knowledge structure comes from all the teaching given to you by the people around you and those that have gone before you.”
He is perhaps suggesting that the advanced systems that we use in our brains point to some of the ways we should be developing AI models in the immediate future. Jepson says that “humans can not compete” with where software engineering is now (in terms of idea and content creation) in AI model creation.
The AI landscape is shifting rapidly – from prompt engineering to adaptive, generative systems that can reconfigure themselves based on context and goals. Jepson’s lightning talk on self-adaptive AI looked at how generative algorithms are changing what it means to build and deploy intelligent systems.
Following a deep data science panel, Jon Knisley presented on the need to stop bleeding out on IT investments. In his role global product marketing leader for Abbyy, Knisley said that unlocking the potential of AI is a critical factor for companies to tackle inefficient processes. With 40% of AI investments wasted due to operations, this session uncovered why process intelligence must be integrated with AI adoption to ensure success.
No AI without PI (process intelligence)
“We have lots of great enterprise data spanning CRM systems and ERP platforms and so on… but what we really don’t have enough of is deep process data so that we can start to apply AI effectively to business processes themselves,” said Knisley. “Given the fact that the average employee switches between poorly integrated applications over 1000 times a day, we have inefficient and workflows inside processes that are in dire need of digital enrichment.”
Knisley made his point by highlighting a quote from May Habib in his position as CEP of Writer.com when she wrote, “You get transformational results from AI when you understand the processes driving the big core workflows that underpin your business. AI can run those workflows only once you understand them. Even with awesome self-orchestrating capabilities, we need blueprints for compliance, for auditability, for improving results.”
Conveying the Abbyy viewpoint here, Knisley said that there is a need to perform process discovery, process analysis, process monitoring, process prediction and process simulation on the road to harnessing AI intelligently in this space.
Delivering on document data
Other sessions at this event included ‘Amplifying AI Potential by Unlocking Your Document Data’ presented by Slavena Hristova. Once a tool for eliminating manual errors, Hristova explains Document AI as now playing a pivotal role in advancing AI-powered automation.
Extended sessions were focused on customer journeys and wider thoughts on the use of intelligent document processing in the future.
In terms of product news emanating from this event, the company has noted that it wants to to solve the pressure software application developers have when attempting to extract reliable and consistent data from business documents. The company introduced ABBYY Document AI, available through a self-service application programming interface (API).
“The ABBYY Document AI API was built with the developer’s experience in mind, allowing users to effortlessly transform unstructured business documents into structured, highly accurate data with just a few lines of code, making it easier to try, integrate, learn and purchase industry leading optical character recognition (OCR) and intelligent document processing (IDP) solutions,” noted the company, in a press statement.
What comes next here…?
Surely our next story here is an assessment of just how far we (and when we say we, we mean Abbyy primarily) have validated the notion of Document AI as a formalised terminology for use in what is clearly a more digitally enabled world of process management that is underpinned by intelligent document processing.
