Exclusive: DocuWare's Marcin Pichur on AI-powered IDP evolution

For decades, businesses have been promised seamless digital workflows, paperless offices and automated data capture.

In reality, organisations still process huge volumes of physical and unstructured documents, and traditional optical character recognition (OCR), despite being mainstream for more than the past couple of decades, has struggled to keep up with the complexity.

Intelligent document processing (IDP) is increasingly being positioned as the breakthrough that finally closes the gap between messy, human-generated paperwork and digital efficiency. For DocuWare, which has spent years building document automation systems around OCR, the arrival of AI and machine learning has fundamentally changed what is possible.

Marcin Pichur, Regional Vice President Sales at DocuWare, describes IDP in simple terms. "In a nutshell, it is the enhanced OCR using state-of-the-art technology that minimises the user intervention necessary," he said. In his view, IDP represents a shift from character recognition to genuine understanding: the system learns layouts, context, and meaning rather than relying on rigid templates.

OCR has existed in various forms for decades, but early systems required painstaking setup. Pichur explained how, while companies once had to create a dedicated template for every document layout: one template for an invoice from an SAP system, a different one for an invoice from Sage, and so on, now that has evolved. 

The limitations were clear. These systems worked only when document volumes justified the labour needed to configure them. A second wave, "free-form recognition," brought more flexibility, but still required configuration and light programming.

Instead of technical staff configuring templates, end users trained the system simply by clicking on fields. Corrections taught the model how to read future documents. "The more people were using the system, the more intelligent it got," Pichur said. Within a few documents, the system could reliably extract information from familiar layouts.

But even that generation had limits. Intelligent indexing worked best with structured documents such as invoices and delivery notes. It struggled with unstructured content like contracts, letters, emails, as well as complex invoices containing multiple tables.

IDP addresses those gaps, said Pichur. These systems can learn from patterns, extract values regardless of where they appear, and eliminate the two biggest investments of older OCR deployments: template configuration and user-based training.

"Regardless of where the information is, the system will learn and will get this information extracted for the end user, which completely eliminates two big investments: The first one is related to configuration of the template, and the second one to the training process by the end users, which can be error prone."

Use cases expanding across industries

While IDP is often associated with accounting, the traditional home of document automation, Pichur argues that far greater value lies elsewhere, in other departments and sectors. Manufacturers, logistics firms, construction companies, universities and insurers are already using the technology to streamline high-volume processes.

Pichur said DocuWare's strongest global customer base remains manufacturing, followed by professional services and then logistics or construction, depending on the region. But Pichur notes that IDP adoption trends across the wider market point to heavy use in finance, healthcare, education and any sector still managing significant paper records.

"These industries…may already have a technology in place that is doing the OCR," he said. But the remaining subset of complex documents often requires a more advanced solution.

In higher education, for example, student applications and onboarding require handling documents in multiple formats: digital files, scans, photographs and handwritten forms. "If we are thinking about the onboarding process of the application process of the students…technologies like IDP can actually help you automate this process," he said.

Insurance claims processing, HR records management, contract administration and supplier onboarding are among the other areas adopting the technology.

Automation is changing roles, not eliminating them

As with most AI-driven tools, IDP raises questions about labour displacement. Pichur acknowledges the concern but says the practical reality is different. New employees, accustomed to mobile apps and AI tools, are often surprised by how manual many business processes still are.

"They started their first work, and they noticed what is happening in reality - paper on the desk, retyping the data into multiple applications," he said. "Did I really study to be doing this job?"

Pichur said fears of widespread job losses have not materialised. In fact, in 14 years at the company, he has seen only one such example, in an African insurance firm with hundreds of data-entry clerks.

In markets like the UK and Germany, he said, the technology has largely shifted staff into higher-value work: customer service, supplier management, contract analysis and other roles that require judgment and interaction.

Asked where IDP is heading by 2026, Pichur said the issue is not a matter of technological capability but organisational readiness. "If you go to any company in the UK, you will see how much paper is still out there," he said.

A DocuWare report from September revealed that 61 per cent of IDP workflows still rely on paper, with 48 per cent of enterprises expecting paper volumes to rise in 2026. 

For Pichur, this underscores the need for incremental but decisive action. The next phase of IDP, then, is less about the leap in AI capabilities and more about businesses' willingness to adopt it.