As a society makes advances, there is usually a recognizable kernel of the past that remains and even forms the basis of ongoing progress. According to Ulf Persson, CEO of ABBYY, documents and automation are inextricably linked. Despite the attention AI is garnering in process automation, much of the most consequential work done by AI happens quietly inside documents people rarely see, he says.
Persson, who sat down with Automation Today recently for a wide-ranging interview, details how decades of experience with documents, processes and regulated customers have shaped ABBYY’s approach to innovation, governance and what comes next as agentic AI finds its way into the enterprise.
Before turning to the future, though, Persson recalled how ABBYY’s market position emerged and why documents remain central to automation efforts today.
Documents are at the center of automation
Persson explains that ABBYY’s origins are firmly rooted in optical character recognition (OCR), and those roots were put down long before AI became a mainstream term. The company initially built and commercialized its own OCR engine, which later became the foundation for broader document capture and understanding capabilities.
“We really came from OCR,” Persson says. “We built our own OCR engine and then realized there was a market not just for the user interface, but for the core engine itself.”
As document capture matured, ABBYY moved beyond supplying components to other vendors and began developing its own capture applications. Persson notes that major shift in the capture market coincided with the rise of robotic process automation, which exposed a structural problem in many automated workflows.
“When RPA came along, people realized that almost every process had documents in the middle of it,” Persson says. “The RPA vendors couldn’t deal with that document dependency, so they came to us.”
Persson explains that while RPA expanded the scope of automation, it also changed expectations. Processes became more transactional, more dynamic, and were increasingly built by business users rather than document specialists. ABBYY, he says, had to rethink how its technology was packaged and delivered.
“We took everything we knew about documents—OCR, classification, extraction—and disassembled it and put it back together again,” Persson says. “That became the foundation for a new platform designed to be modular, flexible and closer to plug-and-play.”
That redesign, he explains, was less about chasing trends and more about adapting to how automation was being deployed inside enterprises.
Innovation driven by customer need
While ABBYY continues to offer new capabilities on its platform, Persson bristles at the idea that innovation is driven by individual features or marketing cycles. Instead, he describes it as a long-term organizational commitment shaped by customer exposure.
“Innovation is really part of your corporate DNA,” Persson says. “You decide to be innovative, or you’re not.”
Persson points to ABBYY’s large installed base of enterprise customers as a key differentiator. According to him, years of production deployments across financial services, insurance and government organizations create constant feedback loops that newer vendors often lack.
“We learn from customers on a daily basis,” Persson says. “Many companies we compete with today simply don’t have that depth of tried-and-tested deployments.”
On the other hand, large technology platforms may have name recognition and excel at building foundational technologies but operate further from real-world document workflows.
“They’re often focused on technology rather than products and solutions,” Persson says. “We hear directly what banks and insurers need, what’s important to them, and what doesn’t work.”
Persson also links innovation to ABBYY’s financial model, noting that long-term profitability has enabled sustained investment in product development without relying on aggressive acquisition strategies.
“We’ve always been profitable and growing,” Persson says. “That allows us to reinvest, but it also forces us to stay at the top of our game.”
Rather than framing innovation as disruption, Persson characterizes it as an iterative process shaped by regulatory environments, operational constraints and customer trust. These factors are often underestimated in discussions about AI adoption, he says.
Governance first, experimentation second
As generative AI becomes embedded in automation platforms, Persson emphasizes that governance concerns now dominate enterprise conversations. He describes regulation, compliance and data security as non-negotiable requirements rather than obstacles to innovation.
“Governance comes up in every single conversation with large customers,” Persson says. “Banks, governments, large enterprises all say it’s top of mind for them.”
Persson acknowledges that adoption has outpaced formal governance structures in many organizations, particularly as AI tools have proliferated rapidly across business units.
“AI came upon us very fast,” Persson says. “People wanted to try it, to dip their feet in, and governance hasn’t always caught up.”
From ABBYY’s perspective, Persson explains, governance has to be built into the platform itself, rather than layered on later.
“For us, it’s trust from the ground up,” Persson says. “Compliance, data security and information security are part of every product release.”
He outlines what he views as practical guidance for enterprises navigating AI adoption: staying ahead of evolving regulations across regions, establishing clear project objectives, maintaining cost controls, and training staff on compliance and guardrails.
“You need clear oversight,” Persson says. “And you need to make sure those guardrails are actually followed inside the organization.”
Persson also highlights the growing importance of security leadership, noting that CISO-to-CISO engagement has become a routine part of enterprise onboarding and vendor evaluation.
“The threat landscape is continuously evolving,” Persson says. “You can’t just say you’re done.”
Agentic AI as an accelerator, not a replacement
Looking ahead, Persson frames agentic AI as an extension of existing automation rather than a radical break from it. He describes agent-based systems as particularly valuable for handling non-linear processes and exceptions that traditional automation struggles to address.
“The biggest difference with agentic AI is that it’s dynamic and interactive,” Persson says. “It can deal with exceptions and back-and-forth in ways that static automation cannot.”
According to Persson, this capability expands the range of processes that can realistically be automated, particularly in customer-facing scenarios such as onboarding, claims processing and account management.
“Many processes are complex and interactive,” Persson says. “Agentic AI makes it possible to automate processes that previously required constant human intervention.”
He cautions, however, against viewing agentic systems as autonomous replacements for enterprise controls. Instead, he positions them as participants in governed workflows, operating within defined constraints.
“An agent can be in the loop,” Persson says. “It can correct, interact and add input. But it must be within guardrails.”
Persson expects the next phase of AI adoption to shift from experimentation to execution, with more large-scale production deployments emerging in the next two years.
“Organizations have learned from early efforts,” he says, “and now we’ll see broader, more successful implementations.”
Rather than predicting a single breakthrough, Persson points to steady expansion, with deeper document intelligence, tighter process integration, and more strategic partnerships as the likely shape of progress ahead.
“It’s about leadership,” Persson says. “Leadership in trust, leadership in innovation, and leadership in how these technologies are applied responsibly.”
As enterprises continue to automate around documents, processes and decisions, Persson believes that the most durable advances come from disciplined execution inside the workflows that already run the business, and that ABBYY will continue to be in the center of it all.
