How Rossum Is Leveraging AI in Document Processing

This December, SSON hosted the 2023 Intelligent Document Processing (IDP) Digital Summit. The online event covered how digitization can streamline document processing, through technologies such as AI.

Within the session “Navigating the Future: Document Processing Automation and AI in Enterprise Transformation”, SSON was fortunately joined by Zach Low. Zach is the Senior Vice President of Sales at Rossum and explained to attendees how the organization uses automation for accurate document processing. In addition, Zach discussed how AI will support Rossum’s vision “to empower one person to handle a million transactions per year.”

The Three V’s

Zach framed the session by outlining three main challenges associated to document processing.

1. Variability: according to the 2022 AFP Payments Fraud and Control Survey, 9% of exchanged documents have discrepancies. This results in approximately $1 million being spent on communications to amend mistakes. These discrepancies can include incorrect banking details, addresses, or sums.
2. Volume: 71% of organizations faced fraud attempts in 2021, with Accounts Payable (AP) being the most targeted sector.
3. Velocity: 53% of new document traffic comes from new business partners annually.

With all this in mind, Zach notes that the number one challenge in document processing is accuracy.

AI In Document Processing

Zach explains how document processing has evolved from manual data entry to digitization, and then optical character recognition (OCR) technology. The industry is now in a new phase of intelligent document processing, which introduces the use of AI. For Zach, this innovative technology “seamlessly moves documents from input through to output at scale with incredibly high accuracy.”

However, Zach posed the question “what is ‘real’ AI, and how does it achieve better accuracy?”

Rossum’s AI

Zach uses Rossum’s AI software to explain the components of ‘real’ AI. There appear to be two main elements: document understanding and end-to-end automation. These two sides cannot exist independently, instead they must collaborate.

On the one hand, Rossum prides itself on its AI’s understanding and “accuracy out-of-the-box”. Their AI has been trained on the countless different transactional documents Rossum has encountered over the last 8 years. The software is “trained [by] every annotation [and] every interaction that happens within the Rossum system.”

On the other hand, true end-to-end automation elevates Rossum’s AI. For example, the AI can spot fraudulent invoices and documents that can often be missed. In addition, Zach highlights that sometimes not all the relevant information is actually written in the document. So, Rossum’s AI can use existing customer data, beyond the document, to make more efficient decisions.

To emphasize these strengths in Rossum’s software, Zach discussed their customer Master Trust Bank of Japan, which saw benefits in just seven weeks.

Overall, Zach urges attendees to embrace continuous innovation. Whether it is through investing in IDP capabilities or leveraging AI, these new technologies will allow organizations to “reclaim valuable manual time, manual effort, enhance[e] productivity and [focus on maximizing accuracy]…”

To learn more about Rossum’s AI, you can access this session on-demand above. Alternatively, to receive more insights into the potential of AI and automation, SSON is thrilled to be hosting the Intelligent Automation Week 2024 in Dallas. Attendees will benefit from insights on overcoming talent gaps, pursuing digital transformation, and the potential of innovative technologies. Secure your place today!