Agentic Performance Management is built on four principles: execution over suggestions, a continuous close that redefines "done," ERP-agnostic and multi-entity by design, and human-in-the-loop governance. Together, they define a new way of running finance, one where agents complete accounting workflows while teams govern outcomes and focus on strategic analysis.
Finance has accumulated a substantial library of automation tools: RPA bots for data entry, AI assistants that draft narratives and surface anomalies, EPM platforms that connect budgeting and reporting workflows, and dedicated tools that track task completion across the close.
After two decades of layering these tools on top of each other, most accounting teams still close the books in roughly the same number of days and manage roughly the same manual workload at period end.
Agentic Performance Management changes how the close gets run. These platforms complete accounting work directly, while teams govern outcomes and review exceptions. That shift rests on four principles, and understanding what they are, and why each one is essential, is the difference between adopting one and actually operating like it.
For leaders evaluating platforms in this space, the label alone reveals little. Vendors describe assistance layers, AI copilots, and basic workflow automation as agentic, and the market has little consistency in how the term gets applied. The four principles that follow give a concrete standard to test against: whether a platform executes on the category's promise or simply borrows its language.
The Four Principles That Define Agentic Performance Management
These principles are architectural decisions, embedded in how a platform is designed instead of positioned in messaging. A platform that embodies all four operates fundamentally differently from one that embodies one or two. They are also interdependent: each one depends on the others to function in a governed, scalable way.
Principle 1: Execution, Not Suggestions
The most consequential distinction in APM is what an agent does when it encounters a reconciliation exception. An assistance-first platform surfaces the exception and recommends a course of action. An agent resolves it, within the parameters it was given, and routes the exception to a human reviewer only when it exceeds defined thresholds.
This is an accounting system that never waits on a human to initiate the next step: agents handle the step, log the outcome, and move forward.
The proof shows up in the numbers. At Team Car Care, the largest Jiffy Lube franchisee in the United States, agents now handle approximately 70% of inventory reconciliation directly, with the remaining 30% surfacing as flagged exceptions for human review.
The shift moved a four-person reconciliation team down to a single agent-supervised role, freeing the accountant who had led that process to move into GL leadership. As CFO Matt Castonguay put it: "It's not just the time saved. It's the confidence of knowing my numbers are right." That result reflects the execution principle at work: a full resolution rather than a smaller efficiency gain layered onto a manual process.
Principle 2: A Continuous Close That Redefines "Done"
Traditional accounting operates in cycles: reconciliation monthly, close at period end, flux analysis only after the numbers are locked. That cycle creates a compression problem: everything has to happen within a few days, which is why most accounting teams spend month-end close in a state of controlled urgency.
Reconciliation and variance review no longer have to wait for a cycle. Agents run them continuously, validating transactions as they post, flagging variances as they arise, and progressing the close throughout the period rather than at its end.
By the time the formal close arrives, most of the work is already complete, and continuous reconciliation across dozens of entities becomes achievable in a way manual processes never allowed, an architectural shift rather than an incremental one.
Jan Grižon, Head of Finance at Leanpay, describes the outcome across five countries: "With Nominal, we've cut at least two days from our close each month, and we save weeks during our end-of-year consolidation." Two days per month reflects a close process that no longer has to catch up at period end because the work was done continuously throughout.
Principle 3: ERP-Agnostic, Multi-Entity by Design
APM sits above the ERP, not inside it. Most finance process automation discussions eventually run into this: organizations have ERPs they have spent years building processes around, and a platform that requires migration will not be adopted at scale. The switching cost and implementation risk are prohibitive.
ERP-agnostic design means the agent layer integrates with whatever the organization already runs, whether NetSuite, SAP, Oracle, Dynamics, or Sage. Multi-entity consolidation follows the same principle: scale doesn't require re-architecting the platform. Whether a company manages five subsidiaries or 280, the same configuration approach applies.
Greg Hood, Head of Finance at Kunai, illustrates the ERP-agnostic side of this principle: "Nominal cut our time to acquisition and saved me an ERP implementation." An ERP implementation that didn't happen represents months of disruption avoided, with the accounting function remaining stable through a deal process.
Principle 4: Human-in-the-Loop Governance
Agentic Performance Management keeps people central to accounting rather than removing them from a regulated function. Human-in-the-loop governance follows from that: defining precisely when and how they're involved, instead of involving them in every execution step.
Every agent action is logged with the rule applied, the data read, and the outcome produced. Human reviewers interact with that log through a three-action interface: approve the outcome, reject an entry that requires revision, or refine the agent's configuration for future runs. Escalations are routed based on pre-defined thresholds. Governance is built into the architecture before deployment, well ahead of any audit finding.
The practical result is that accounting teams spend their time reviewing what agents completed and what they escalated. Josh Ramos at GSPP describes what that looks like in practice: "Nominal consolidated everything with a click, saving us countless hours and keeping our books audit-ready." That readiness becomes a standard output rather than a separate prep effort, the natural result of governance embedded in execution from the start.
Want the full breakdown of all four principles, with implementation benchmarks and a platform evaluation checklist? Download the Agentic Performance Management Playbook.
Why These Principles Matter Together
Each principle is meaningful individually. Together, they create the operating model that makes autonomous accounting viable at scale. Execution without governance is a liability. Governance without execution is a more organized version of the problem accounting teams already have. Continuous operation that still requires ERP replacement will not get deployed. The four principles are built to operate together.
Where Other Automation Approaches Fall Short
The gap between APM and other automation categories becomes clear when the four principles are applied as evaluation criteria. RPA executes but doesn't operate continuously and lacks a governance model designed for accounting compliance. AI assistants are continuous and reviewable, but they assist rather than execute. EPM tools provide planning and governance capability without an agentic workflow execution layer.
This is the category in which all four principles operate simultaneously. Agents execute accounting work continuously, within an ERP-agnostic architecture, governed by human oversight at defined decision points. Partial implementations, in which one or two principles are active but others are not, preserve the manual workload in the gaps the platform doesn't cover.
What Finance Looks Like When All Four Are Active
In an organization operating on all four principles, the close process looks like this: agents execute reconciliations continuously throughout the period, flag exceptions to the appropriate reviewer, complete intercompany eliminations as intercompany transactions post, and produce close documentation as a byproduct of normal execution.
The accounting team reviews a summary of what was completed, approves the period-end close, and turns its attention to analysis and planning. Close becomes a confirmation of work already done, without the last-minute sprint to finish it.
Applying the Principles in Practice
Implementing APM doesn't require activating all four principles at once across every workflow. It means starting with high-confidence, high-volume processes and building governance infrastructure before expanding into more complex territory.
Start With Execution in Repeatable Workflows
Bank reconciliation, standard journal entries, and subledger-to-GL matching are the natural entry points for Agentic Performance Management deployment. These workflows are high volume, well-defined, and fast to configure. The gains are measurable within the first close cycle.
Build Governance Before Expanding Scope
Before expanding into more complex workflows, including multi-entity consolidation and intercompany eliminations, define the governance model in detail: who reviews exceptions, what thresholds trigger escalation, and how actions are logged and approved. Governance infrastructure built for simple workflows applies consistently as scope expands.
Measure the Close, Not the Technology
The metric that matters is days to close, reconciliation backlog, and hours spent on manual entries. The number of agents deployed or workflows configured is not an outcome measure. APM should produce measurable reductions in close cycle length and manual accounting effort, typically within the first quarter of deployment.
Agentic Performance Management is defined by these four principles, and any platform claiming to represent the category should be evaluated against all four. Nominal was built around this complete set, executing reconciliations, eliminations, and close workflows under human governance instead of layering partial automation onto a manual process.
Teams that deploy platforms embodying all four will close faster, report with greater confidence, and scale accounting coverage without scaling headcount proportionally. See how Nominal puts all four principles to work in your close: book a demo.
About the writer
