At Sibos 2023 in Toronto, John Bevil, senior product manager, Xceptor, and John Joseph, head of North American sales engineering, Xceptor, chat with Funds Europe about the challenges and innovations in data reconciliations, emphasising the balance between rapid onboarding and maintaining accuracy.
Funds Europe: The capital markets are inundated with diverse unstructured data. Based on your experience, which types or data sources have posed the most significant challenges for businesses regarding reconciliations, and what are the reasons for these challenges?
John Bevil: From what I’ve observed, the most intricate data types tend to be those that don’t conform to the standard infrastructure – data that’s not part of the core. Typical examples include PDFs and Excel spreadsheets. These formats, and others that aren’t standardised, often exist outside of the main data hubs or “centres of excellence” in a business. Consequently, they’re found on the peripheries, which makes them a primary focus of concern when dealing with unstructured data.
John Joseph: Building on John’s mention of PDFs, it’s the more intricate documents like credit agreements and lengthy loan notices where data extraction becomes complex. Imagine extracting specific data from scattered pages in a 150-page document. Then there’s the challenge of emails. For instance, a significant portion of standing settlement instructions is still communicated via email. Automating the extraction of critical data points from such emails, especially when embedded in the body text, presents considerable challenges.
Funds Europe: AI’s capacity to comprehend and structure data is becoming pivotal in reconciliations. Given the dynamic nature of data formats in the industry, how do you envision the evolution of AI in this area?
John Joseph: It’s a multifaceted issue. While AI can extract data from unstructured sources and assert a confidence level, say 95%, it’s not foolproof. Pushing such data into reconciliations poses challenges. The key is to harness AI efficiently, set clear boundaries, and follow a rigorous post-extraction process to enhance and rectify any discrepancies. Additionally, AI can offer advanced matching criteria for reconciliations, which is beneficial.
John Bevil: I’ve frequently heard discussions on AI throughout the week, and a recurring theme is that the effectiveness of AI is tightly bound to the quality of the data it receives. AI will have a more significant role in reconciliations as we move forward. Still, it’s crucial to remain aware of the regulatory landscape and the potential impacts on a company’s bottom line.
Funds Europe: Rapid onboarding and setup are now industry imperatives. Is there a danger that the urgency for speed might undermine the accuracy or reliability of reconciliations? How can companies strike the right balance?
John Joseph: Why not optimise for both speed and accuracy? For instance, Xceptor has a client who drastically reduced their onboarding time while improving efficiency and minimising errors. The key lies in establishing repeatable, well-defined processes that ensure consistent onboarding quality while accelerating the process.
Funds Europe: With technological advancements driving automation, do you foresee a time when manual reconciliations might be rendered obsolete? What would this paradigm shift mean for roles traditionally centred around these tasks?
John Bevil: I genuinely believe we’re heading towards a significant transformation in manual reconciliations. The inherent data challenges and the evolution of “centres of excellence” in organisations catalyse this change. These centres aim to consolidate and streamline reconciliations, and some platforms play a crucial role in this evolution. Over time, as more manual tasks are automated and integrated into these centres, the very nature of manual reconciliations is bound to change.
John Joseph: While automation is vital for efficiency, I have a particular concern: we should never lose sight of the data’s integrity. Bots or AI solutions replacing manual reconciliations can inadvertently obscure underlying data issues. What’s essential is that we create institutional processes that ensure data quality and integrity, focusing on the data rather than just automating tasks for efficiency or cost savings.
© 2023 funds europe
