Agentic AI
Autonomous Finance
AI
The Role of AI Agents in Intercompany Reconciliation: An Interview with EY Advisors
Uncover insights from EY advisors on the role of Agentic AI in intercompany reconciliation.
Q&A with
Following a recent webinar on the topic of automating intercompany reconciliation, I met with Andreas Muzzu and Vish Dhingra from EY to have a deeper dialogue on this topic. We explored the challenges and discussed how the use of AI and document intelligence systems like Hypatos can automate and improve non value adding financial processes like intercompany reconciliation.
Ana: Thank you, Andreas and Vish, for sharing your insights on this challenging topic. Having followed the evolution of the financial function over years, between the two of you, you have a wealth of knowledge and experience. Andreas, at EY you’re managing a client-centric team supporting CFOs. How do you see the role of the CFO evolving?
Andreas: I’d like to highlight a recent survey that examines the DNA of CFOs conducted by EY. We see the role evolving into what is called the CVO – the Chief Value Officer. Today, CFOs face complex and contradictory demands. They’re under pressure to create long term value. They have to manage risks while finding more innovative ways to create value within finance, and lastly, they are struggling to act as strategic finance leaders with traditional skillsets that often don’t match current needs.
Ana: What are some ways CFOs can focus on value topics?
Andreas Muzzu: One way is to eliminate non-value-adding transactional tasks like managing Accounts payable, Accounts receivabe or intercompany reconciliation that require a lot of resources. This is where AI can be of great value. It can do the repetitive, time-consuming tasks that are currently done by humans at less cost and effort. We know of companies that have up to 20 million unreconciled items that have added up over the years at great expense.
Ana: Vish, can you explain your role as Global Transformation Leader for Financial Advisory Accounting Services (FAAS) and how you are advising finance departments to address this issue.
Vish Dhingra: I help companies transform their finance departments into higher value adding entities. Clearly, reconciling differences between sales orders and purchase orders is not a value-adding activity, but it must be performed to avoid issues with tax authorities and other regulatory entities. Besides taxation, there are numerous issues when it comes to intercompany reconciliation such as different pricing in different countries, different currencies, and data discrepancies. A purchase order issued in one country could be for 110 items, for example, while the sales order corresponding to that purchase in another country could be for 100. And at delivery, maybe 5 items are found to be defective and need replacement.
Ana: What are some solutions to this dilemma?
Vish: Automating the process helps, but there is a long way to go. The best solution would be to get to the root cause and avoid the need for reconciliation by matching debit and credit transactions on day one. In an ideal world, the mismatch would be avoided if the sales order and purchase order were issued simultaneously.
Andreas: That would be easy if you only had one SAP instance, for example. However, one of the main problems is that many multinationals are dealing with different ERP systems. This is partly due to acquisitions with legacy systems or because systems are upgraded or managed differently in different parts of a company. Plus you should look at one aspect: If you do not have access to the underlying and guiding documents of a transaction (e.g. Purchase order, purchase agreement, Accounting and Tax policies, Freight documents etc.) you will only be able to reconcile based on the recorded data in the system. Agentic AI can provide matching based on structured and unstructured data. At one point, it was thought that potentially blockchain would help resolve data inconsistencies, but that did not materialize. Now, it’s clear that agentic AI can play a big role. But let’s not forget that implementing AI in this capacity is not a plug and play exercise. It requires a great deal of customization.
Ana: What are some of the key success factors?
Vish: I think it starts with the underlying data. Even if you have the best AI models, if the data is not good, the results won’t be good either. And cost is still a big factor. In some cases, we see companies trying to solve reconciliation issues with a lot of people doing the work manually in low-cost countries, instead of deploying expensive technology, such as AI.
Andreas: Many people are very excited about AI. They want to apply it, but they don’t really know where to start. It requires trained people, but there aren’t enough trained people. There’s also a great deal of concern about making people redundant, especially in accounts payable departments. Managers are reluctant to replace people with the tools because if they did, they would either have to let people go or find value-adding tasks for them.
Vish: What it really means is that the accountants currently doing intercompany reconciliations would have to learn how to communicate with the AI, learn how to tell the AI what to do, and then review the outcome of the work. And that’s not easy. It requires more skills. This applies to us as well. As consultants our roles are also changing. We are still the subject matter experts, but we must learn how to work with AI as we would with any new employees.
Andreas: For example, we believe that starting an AI implementation should be tackled in the same way as an ERP implementation project. Start by establishing clear goals and objectives for what you want to achieve, with a detailed project scope and engagement plan. Next, as with any ERP implementation, data quality is crucial. Low-quality data can have serious repercussions on financial processes, leading to inaccuracies that harm businesses. AI algorithms are powerful tools, but their performance is strictly related to the information they receive. And lastly, you need to assemble a skilled project team of trained experts and develop a clear change management plan.
Ana: How can AI tools like the ones offered by Hypatos help fill the gaps so CFOs can start delivering more value with non-transactional activities?
Vish: In a nutshell, to start delivering more value you need to bring together people, process, and technology. You need to use clean data and implement governance for auditable results. In finance, there are basically two key processes – Order-to-Cash (O2C) and Procure-to-Pay (P2P). Hypatos offers AI-powered automation solutions for document processing that can significantly enhance efficiency and accuracy in both processes.
Andreas: Agentic AI’ solutions should be designed to integrate with ERP systems like SAP, allowing businesses to enhance their O2C and P2P workflows without replacing existing systems. By automating these repetitive, document-heavy tasks, Hypatos helps finance teams save time, reduce costs, and improve process accuracy, leading to more efficient and compliant operations.
Ana: Thank you, Andreas and Vish!
To learn more about this topic:
Further stories from our blog
.webp)
