As finance organizations accelerate AI adoption, many are discovering that traditional automation and AI copilots still leave some of the most critical finance processes dependent on manual effort. While ERP systems have become the system of record for enterprise finance, activities such as reconciliations, intercompany accounting, variance analysis, and financial close often continue to rely on spreadsheets and human intervention. The next phase of enterprise AI is shifting the focus from assisting finance professionals to executing complex financial workflows with governance, auditability, and human oversight built in.
In this ERP News Executive Q&A, Guy Leibovitz, Co-Founder and CEO of Nominal, discusses why finance teams need more than AI copilots, how autonomous AI agents are reshaping financial operations, and why the future of ERP lies in an execution layer that works alongside existing enterprise systems. He also shares his perspective on continuous close, governance, and what will distinguish finance organizations that successfully scale with AI over the coming years.
From Finance Automation to Autonomous Execution
Q: Why are finance teams still struggling with manual work despite significant investment in ERP systems, automation, and AI?
A: Because most of that investment went into recording and reporting, not execution. ERPs capture the data. Automation moves it from one screen to another. But the actual work, matching transactions, reconciling intercompany activity, explaining variances, still sits with a person and a spreadsheet. You can buy every tool on the market and still close the books by hand if nothing is actually executing the workflow end to end.
Q: What are the biggest limitations of today’s AI copilots in finance environments?
A: Copilots answer questions. They don’t do the work. Ask a copilot why intercompany is out of balance and it might tell you. It won’t go fix it. That’s the line between assistive AI and execution. A copilot speeds up a human’s task. It doesn’t remove the task. Finance teams don’t need a faster way to look at the problem. They need the problem handled.
“Copilots answer questions. They don’t do the work. That’s the line between assistive AI and execution.”
Q: Many vendors are racing to launch AI agents and AI copilots. What should CFOs and finance leaders be looking for when evaluating whether these technologies can deliver real business value?
A: Ask one question: does this system execute a workflow on its own, start to finish, or does it just help a person execute it faster? Everyone calls everything “agentic” right now. The test is simple. Can it run a reconciliation, flag what needs a human, and close the loop without someone driving every step? If the answer is no, you’re buying a better interface, not a new capability.
Rethinking the Close for Multi-Entity Finance
Q: As companies grow across entities, geographies, and business units, complexity often scales faster than revenue. What are the biggest operational challenges multi-entity finance teams face today?
A: Every new entity multiplies the close. More intercompany transactions, more currencies, more local rules, more reconciliations that all have to land before consolidation can even start. Teams scale headcount to keep up, but headcount doesn’t scale as fast as entity count. The work grows geometrically. The team grows linearly, if it grows at all.
Q: Why has the concept of a continuous close remained elusive for so many organizations?
A: Because continuous close requires continuous execution, and most finance stacks are built for periodic execution. People run the close once a month because that’s when someone sits down and does the matching, the reconciling, the reviewing. If the work only happens when a person does it, the close can only ever be as continuous as that person’s calendar. Agents change that math. They don’t wait for month-end to start working.
Q: Which finance processes are genuinely ready for autonomous execution today, and which still require significant human judgment and oversight?
A: Bank reconciliation, intercompany matching, transaction classification, and flux explanation are ready now. These are high-volume, rules-based, pattern-heavy processes. Judgment calls, policy decisions, anything with material risk or ambiguity, that stays with the controller or the CFO. The agents execute the workflow. The human owns the decision. That split doesn’t move. Ten years from now, finance teams won’t measure success by how quickly they close the books. They’ll measure success by how little work is left to do at month-end. That’s the shift AI makes possible.
Autonomy Still Requires Accountability
Q: How should finance leaders think about governance, auditability, and accountability when introducing AI into core financial operations?
A: Every action an agent takes needs a record. Not a summary, the actual decision trail: what it matched, why, against what rule, with what confidence. If you can’t audit an agent’s work the same way you’d audit a junior accountant’s, it doesn’t belong in your close. Human-in-the-loop isn’t a nice-to-have. It’s the only way autonomy and accountability coexist.
“If you can’t audit an agent’s work the same way you’d audit a junior accountant’s, it doesn’t belong in your close.”
Q: What role does organizational knowledge and historical context play in making AI effective inside finance teams?
A: It’s the difference between an agent that executes correctly and one that executes confidently but wrong. Every company has its own exceptions, its own historical quirks in the chart of accounts, its own reasons a certain intercompany balance never quite nets to zero. An agent that doesn’t carry that context will flag noise as signal constantly.
The ones that work are the ones trained on the company’s actual financial history, not a generic model of what finance should look like.
ERP as the System of Record, AI as the Execution Layer
Q: ERP systems have long served as the system of record for enterprise operations and finance. As AI becomes more capable of analyzing information and executing tasks, how do you see the relationship between ERP platforms, finance systems, and AI evolving over the next few years?
A: The ERP stays the system of record. That doesn’t change. What changes is everything that happens on top of it. Today most of that layer is manual or semi-automated. Over the next few years it becomes an execution layer, agents that read the ERP, run the close, reconcile across entities, and hand back clean books. NetSuite, SAP, Oracle, Workday, Sage, they hold the data. The execution layer is where the work actually gets done.
Q: Looking ahead, what will distinguish the finance organizations that successfully scale with AI from those that simply automate existing inefficiencies?
A: The ones that scale are the ones that rebuilt the workflow around execution instead of bolting AI onto the process they already had. If you automate a broken process, you get a faster broken process. The teams that win are running their close, their reconciliations, their consolidations through agents that actually execute the work, with people reviewing outcomes instead of producing them. That’s the real shift. Not faster automation. A different model entirely.
ERP News Editorial Team
The ERPNews Editorial Team covers global developments in ERP (Enterprise Resource Planning), enterprise software, cloud platforms, AI, automation, and digital transformation, providing independent news and editorial analysis for senior business and technology leaders. Our reporting focuses on market signals, strategic shifts, and enterprise impact across the ERP and enterprise technology ecosystem.
For editorial inquiries, please contact:
📩 [email protected]
