Explore how AI agents are revolutionizing intercompany reconciliation, turning reactive fixes into proactive solutions and transforming financial operations worldwide.
During our recent visit to São Paulo and Mexico City, where we engaged with various industry leaders, one conversation particularly resonated - the universal challenge of intercompany reconciliation. While discussing innovation in finance with Brazilian executives, their experiences echoed a recent encounter we had with a global Oil & Gas company that perfectly illustrates the industry-wide challenge we're facing.
This Oil & Gas giant was grappling with millions of unreconciled intercompany transactions, accumulated over several reporting periods. Picture this: transactions posted months or even years ago, each lacking proper cross-validation with their counter entries. The result? A massive backlog of mismatches in prices, deliveries, and FX rates required extensive investigation long after the fact.
The challenge was exponentially complicated by the time lag. Trying to investigate transaction details from years past proved to be a logistical nightmare:
The costs and risks associated with such reconciliation backlogs create a cascade of business challenges:
Current reconciliation solutions typically focus on matching transactions after they occur. While these tools have brought some improvement, they face several limitations:
Our new approach developed together with EY FAAS Global Innovation & Digital, leveraging AI agents, fundamentally transforms intercompany reconciliation from reactive to proactive.
This process flow illustrates an intercompany transaction system enhanced by an Agentic AI solution. The "Human in the Loop” still performs important tasks and interacts with several AI agents. The AI agents play a key role in automating and verifying transactions across multiple ERP systems, facilitating efficient intercompany reconciliation by reducing manual effort and ensuring compliance. The clear advantage of an AI agent over traditional matching/reconciliation software derives from “understanding” content in the underlying documents and matching this content with various other relevant documents across the entities, e.g. General Terms and conditions, Finance Policies, Freight documents, Purchase orders etc. This process is called “multi-level” matching. Here's a step-by-step breakdown of the process for an example supply chain case (sale of material):
Process Improvements
The system operates on a holistic approach that does not rely on a monolithic framework across all group companies but is capable of matching and reconciling diverse data sources and formats used across multiple ERPs. Automated validation and compliance are integrated at every step of the process, with manual intervention limited to cases of true matching differences or exceptions that require human resolution. An AI agent autonomously addresses such discrepancies by communicating with counterparties responsible for issues, such as those arising from inventory limitations, damages, or updated price lists. Real-time processing enables the system to function fully autonomously, with standardized handling of all transactions ensuring process autonomy. Built-in controls resolve matching differences promptly while maintaining compliance with laws and regulations. The system incorporates clear activity separation and provides end-to-end visibility to enhance efficiency and transparency.
Centralized Control and Reporting
Through the Hypatos Insights application, organizations gain:
Business Value
The LATAM connection
Our discussions in EY’s São Paulo office during a “Future of Finance” event revealed how universal these challenges are. Whether it's a Brazilian conglomerate managing multiple subsidiaries or a global Oil & Gas company, the need for proactive reconciliation is clear. The enthusiasm among Brazilian finance leaders for this AI-driven approach reinforced that we're on the right track.
Watch our webinar on demand to learn how to transform your intercompany reconciliation from a million-dollar headache into a streamlined, automated process.
Further stories from our blog