Teams researching AI-driven finance operations run into a growing set of terms that sound similar but mean very different things. This glossary defines the vocabulary of Agentic Performance Management, from the agents that execute accounting workflows to the governance models that keep them auditable.
Use it as a reference when evaluating what a term actually means for your close process, not just what it sounds like it means.
An intelligent, coordinated system of specialized AI agents that operates on top of existing ERPs to independently execute multi-step finance and accounting workflows, with audit-ready traceability and human oversight. Unlike tools that track or suggest, Agentic Performance Management platforms execute accounting workflows and report outcomes.
An AI system that continuously monitors financial data to identify errors, irregularities, and policy violations before they affect the close or trigger audit findings. Unlike rules-based tools that only catch predefined conditions, anomaly detection agents read transaction content and historical patterns to flag issues that no rule was written to anticipate. Nominal implements this capability through Transaction patrol agents.
AI systems designed to act autonomously over multi-step tasks, adapting to context and maintaining state across actions. Agentic AI is the underlying capability that enables APM platforms, allowing agents to handle accounting workflows from trigger to completion without constant human direction.
A software system that perceives its environment, makes decisions, and takes actions to achieve a defined goal without requiring human input for each step. In accounting, AI agents execute workflows, including reconciliation, consolidation, and close, that previously required manual effort at every stage.
A software agent that completes multi-step workflows without continuous human instruction. In accounting, autonomous agents apply pre-configured logic to reconcile transactions, prepare journal entries, perform eliminations, and escalate exceptions, operating continuously rather than only at period end.
A category of software that tracks how well a company is performing against its strategic goals, producing outputs like balanced scorecards and KPI dashboards. BPM reports on outcomes rather than producing them. Where BPM tracks and EPM plans, APM executes the underlying accounting work both depend on.
The process of matching internal cash records against bank statement activity to confirm that recorded transactions and actual bank activity agree. Traditionally a month-end task performed manually, bank reconciliation is one of the most time-consuming recurring controls in the close.
An agent that continuously compares bank feed activity against ledger data, matching transactions as they occur rather than in a single month-end batch. It surfaces missing entries, duplicate payments, and unexpected fees as they arise, keeping cash accounts reconciled throughout the month instead of only after close.
An agent that oversees the overall close process, tracking task completion, escalating blockers, coordinating agent and human activity, and producing close documentation. Close management agents replace manual checklist coordination with continuous operational oversight.
An agent that executes the financial consolidation process for multi-entity organizations, handling intercompany eliminations, currency translation, and consolidated balance preparation. Consolidation agents operate at the entity level, progressing the close for each subsidiary on a continuous basis.
A second, independent review step that checks an agent's output before it reaches a human. If a generated journal entry has debits that do not equal credits, or an anomaly flag does not hold up under scrutiny, the critic layer blocks or filters it. This ensures only validated recommendations reach the finance team for review.
A close model in which accounting agents validate transactions, post journal entries, and progress the close on a rolling basis rather than compressing all activity into a period-end sprint. Continuous close eliminates the month-end crunch by keeping the books current throughout the period.
An intelligent, coordinated system of specialized AI agents that operates on top of existing ERPs to independently execute multi-step finance and accounting workflows, with audit-ready traceability and human oversight. Unlike tools that track or suggest, Agentic Performance Management platforms execute accounting workflows and report outcomes.
An AI system that continuously monitors financial data to identify errors, irregularities, and policy violations before they affect the close or trigger audit findings. Unlike rules-based tools that only catch predefined conditions, anomaly detection agents read transaction content and historical patterns to flag issues that no rule was written to anticipate. Nominal implements this capability through Transaction patrol agents.
AI systems designed to act autonomously over multi-step tasks, adapting to context and maintaining state across actions. Agentic AI is the underlying capability that enables APM platforms, allowing agents to handle accounting workflows from trigger to completion without constant human direction.
A software system that perceives its environment, makes decisions, and takes actions to achieve a defined goal without requiring human input for each step. In accounting, AI agents execute workflows, including reconciliation, consolidation, and close, that previously required manual effort at every stage.
A software agent that completes multi-step workflows without continuous human instruction. In accounting, autonomous agents apply pre-configured logic to reconcile transactions, prepare journal entries, perform eliminations, and escalate exceptions, operating continuously rather than only at period end.
A category of software that tracks how well a company is performing against its strategic goals, producing outputs like balanced scorecards and KPI dashboards. BPM reports on outcomes rather than producing them. Where BPM tracks and EPM plans, APM executes the underlying accounting work both depend on.
The process of matching internal cash records against bank statement activity to confirm that recorded transactions and actual bank activity agree. Traditionally a month-end task performed manually, bank reconciliation is one of the most time-consuming recurring controls in the close.
An agent that continuously compares bank feed activity against ledger data, matching transactions as they occur rather than in a single month-end batch. It surfaces missing entries, duplicate payments, and unexpected fees as they arise, keeping cash accounts reconciled throughout the month instead of only after close.
An agent that oversees the overall close process, tracking task completion, escalating blockers, coordinating agent and human activity, and producing close documentation. Close management agents replace manual checklist coordination with continuous operational oversight.
An agent that executes the financial consolidation process for multi-entity organizations, handling intercompany eliminations, currency translation, and consolidated balance preparation. Consolidation agents operate at the entity level, progressing the close for each subsidiary on a continuous basis.
A second, independent review step that checks an agent's output before it reaches a human. If a generated journal entry has debits that do not equal credits, or an anomaly flag does not hold up under scrutiny, the critic layer blocks or filters it. This ensures only validated recommendations reach the finance team for review.
A close model in which accounting agents validate transactions, post journal entries, and progress the close on a rolling basis rather than compressing all activity into a period-end sprint. Continuous close eliminates the month-end crunch by keeping the books current throughout the period.
An agent that applies fixed, pre-approved logic to produce consistent, reproducible outputs. Deterministic agents are essential in accounting because every action must be explainable and auditable. Given the same input, a deterministic agent always produces the same output, which is critical for SOX compliance and audit trail integrity.
Software used for financial planning, budgeting, forecasting, and reporting. EPM tools are planning systems that model what finance teams expect to happen. APM is the execution layer that handles what actually happens in the books. In short: EPM plans and APM executes.
The system of record for financial transactions, including the general ledger, accounts payable, accounts receivable, and subledger data. ERPs record and store financial data but do not execute accounting workflows autonomously. APM operates on top of the ERP, extending its capability without replacing it.
A design principle meaning the APM layer integrates with any existing ERP (NetSuite, SAP, Oracle, Sage, and others) without requiring system replacement. ERP-agnostic platforms are additive, extending the capability of existing systems rather than disrupting them.
The end-to-end process from the last day of an accounting period to the publication of final financial statements, including reconciliation, journal entries, consolidation, review, and sign-off. Reducing the financial close cycle from 10 to 15 days down to 3 to 5 days is a primary outcome of APM adoption.
An agent that identifies and explains period-over-period variances in financial results. Flux agents compare current and prior period balances, flag movements above defined thresholds, and generate explanations grounded in transaction-level data, replacing manual flux analysis narratives. Also called a variance analysis agent.
The analysis of period-over-period changes in account balances, identifying the transactions or conditions that explain movements. Flux analysis is a standard audit and management reporting requirement. In APM, flux agents perform this analysis automatically and produce explanations tied to source data.
An agent that applies fixed, pre-approved logic to produce consistent, reproducible outputs. Deterministic agents are essential in accounting because every action must be explainable and auditable. Given the same input, a deterministic agent always produces the same output, which is critical for SOX compliance and audit trail integrity.
Software used for financial planning, budgeting, forecasting, and reporting. EPM tools are planning systems that model what finance teams expect to happen. APM is the execution layer that handles what actually happens in the books. In short: EPM plans and APM executes.
The system of record for financial transactions, including the general ledger, accounts payable, accounts receivable, and subledger data. ERPs record and store financial data but do not execute accounting workflows autonomously. APM operates on top of the ERP, extending its capability without replacing it.
A design principle meaning the APM layer integrates with any existing ERP (NetSuite, SAP, Oracle, Sage, and others) without requiring system replacement. ERP-agnostic platforms are additive, extending the capability of existing systems rather than disrupting them.
The end-to-end process from the last day of an accounting period to the publication of final financial statements, including reconciliation, journal entries, consolidation, review, and sign-off. Reducing the financial close cycle from 10 to 15 days down to 3 to 5 days is a primary outcome of APM adoption.
An agent that identifies and explains period-over-period variances in financial results. Flux agents compare current and prior period balances, flag movements above defined thresholds, and generate explanations grounded in transaction-level data, replacing manual flux analysis narratives. Also called a variance analysis agent.
The analysis of period-over-period changes in account balances, identifying the transactions or conditions that explain movements. Flux analysis is a standard audit and management reporting requirement. In APM, flux agents perform this analysis automatically and produce explanations tied to source data.
A design principle describing agents that operate directly on accounting data structures, accessing accounts, entities, dimensions, and transaction memos rather than working from summary balances. GL-native agents understand that debits must equal credits and that intercompany transactions require elimination entries, giving them the context to validate their own outputs.
A governance model where agents execute accounting tasks autonomously, and humans review and approve outcomes at defined checkpoints. Human-in-the-loop governance is what makes APM compliant: agents act within defined parameters, exceptions are escalated, and every action is logged and reviewable.
The accounting process of removing intercompany transactions, such as loans, management fees, and sales, from consolidated financial statements to prevent double-counting. Elimination entries offset intercompany balances across entities. In multi-entity organizations, elimination is one of the most complex and error-prone manual accounting tasks.
The process of verifying that intercompany balances recorded by two related entities match before consolidation. Unreconciled balances are a primary cause of consolidation delays and audit findings. Agents handle intercompany reconciliation continuously rather than at period end.
A design principle describing agents that operate directly on accounting data structures, accessing accounts, entities, dimensions, and transaction memos rather than working from summary balances. GL-native agents understand that debits must equal credits and that intercompany transactions require elimination entries, giving them the context to validate their own outputs.
A governance model where agents execute accounting tasks autonomously, and humans review and approve outcomes at defined checkpoints. Human-in-the-loop governance is what makes APM compliant: agents act within defined parameters, exceptions are escalated, and every action is logged and reviewable.
The accounting process of removing intercompany transactions, such as loans, management fees, and sales, from consolidated financial statements to prevent double-counting. Elimination entries offset intercompany balances across entities. In multi-entity organizations, elimination is one of the most complex and error-prone manual accounting tasks.
The process of verifying that intercompany balances recorded by two related entities match before consolidation. Unreconciled balances are a primary cause of consolidation delays and audit findings. Agents handle intercompany reconciliation continuously rather than at period end.
A record of a financial transaction posted to the general ledger, including the accounts debited and credited, the amount, and the period. In APM, agents post journal entries automatically for recurring transactions, corrections, and eliminations, with human review for exceptions.
An agent that performs transaction-level matching between source data sets, such as invoices to payments, intercompany transactions between entities, or subledger entries to general ledger postings. Matching agents are the foundation of high-volume reconciliation in multi-entity or high-transaction environments.
A specialized anomaly detection agent that identifies transactions posted to the wrong account or dimension, such as a laptop purchase booked to software expenses. It works by reading transaction descriptions and comparing them against expected accounting logic, catching errors that rules-based checks cannot.
A specialized anomaly detection agent that identifies recurring transactions that unexpectedly stop appearing, such as a vendor payment that has posted monthly for a year and then does not. It recognizes the pattern rather than checking for a single missing entry, including when the missing amount is spread across multiple line items.
The process of finalizing financial records for a given accounting period, including reconciliations, journal entries, eliminations, and financial statement preparation. In traditional environments, this is a compressed, labor-intensive sprint. Under APM, it becomes a continuous process with a formal sign-off at period end.
The process of combining financial results from multiple legal entities into a single consolidated financial statement, after eliminating intercompany transactions and applying currency translations. Multi-entity consolidation becomes significantly more complex as entity count and intercompany relationship volume increase.
The ability for finance teams to define agent logic and thresholds in plain language rather than code or rigid rule engines, for example "reverse all matched lines and post the balancing amount to account 5100." This lets controllers and accounting managers create, adjust, and refine agent behavior without engineering support.
A record of a financial transaction posted to the general ledger, including the accounts debited and credited, the amount, and the period. In APM, agents post journal entries automatically for recurring transactions, corrections, and eliminations, with human review for exceptions.
An agent that performs transaction-level matching between source data sets, such as invoices to payments, intercompany transactions between entities, or subledger entries to general ledger postings. Matching agents are the foundation of high-volume reconciliation in multi-entity or high-transaction environments.
A specialized anomaly detection agent that identifies transactions posted to the wrong account or dimension, such as a laptop purchase booked to software expenses. It works by reading transaction descriptions and comparing them against expected accounting logic, catching errors that rules-based checks cannot.
A specialized anomaly detection agent that identifies recurring transactions that unexpectedly stop appearing, such as a vendor payment that has posted monthly for a year and then does not. It recognizes the pattern rather than checking for a single missing entry, including when the missing amount is spread across multiple line items.
The process of finalizing financial records for a given accounting period, including reconciliations, journal entries, eliminations, and financial statement preparation. In traditional environments, this is a compressed, labor-intensive sprint. Under APM, it becomes a continuous process with a formal sign-off at period end.
The process of combining financial results from multiple legal entities into a single consolidated financial statement, after eliminating intercompany transactions and applying currency translations. Multi-entity consolidation becomes significantly more complex as entity count and intercompany relationship volume increase.
The ability for finance teams to define agent logic and thresholds in plain language rather than code or rigid rule engines, for example "reverse all matched lines and post the balancing amount to account 5100." This lets controllers and accounting managers create, adjust, and refine agent behavior without engineering support.
An agent that matches transactions across accounts, systems, or entities, applying defined matching logic to identify matches, flag discrepancies, and post correcting entries. Reconciliation agents handle bank reconciliations, subledger-to-GL reconciliations, and account balance verifications.
An agent that automatically generates corrective journal entries to fix discrepancies identified during reconciliation, such as an unmatched transaction flagged by a matching agent or an anomaly surfaced by transaction patrol. Resolution agents complete the loop between detecting an issue and fixing it, producing a properly structured, auditable entry for human review rather than leaving the correction to manual work.
Automation technology that replicates human interactions with software interfaces, such as clicking, copying, and pasting, to execute repeatable tasks. RPA works in stable environments but breaks when data formats or systems change. Unlike agentic AI, RPA cannot interpret context, handle exceptions, or adapt to variation.
A complete, transaction-level copy of the general ledger that Nominal maintains alongside the customer's ERP, including memos, dimensions, and historical patterns. Most finance automation tools work only with summary balances pulled from the ERP; a shadow general ledger gives agents the detail needed to trace a variance back to the specific transactions that caused it.
The oversight layer within an APM platform that assigns agent-generated recommendations to the appropriate reviewer and tracks approval status. Nothing posts to the ERP until a human explicitly signs off, making task management the mechanism that keeps agent execution human-governed rather than autonomous.
The combination of three platform components: a native general ledger, comprehensive task management, and embedded AI agents. Together, let an APM platform execute complete workflows rather than just one piece of them. Competitors typically offer one or two of these components; Nominal's model depends on having all three working together.
The process of pairing individual transactions from two or more data sources, for example, matching a payment to an invoice or an intercompany transaction to its counterpart in another entity's records. Transaction matching at scale is one of the primary use cases for matching agents.
An agent that monitors transactions continuously for anomalies, errors, and compliance exceptions — flagging issues before they reach the close rather than discovering them during reconciliation. Patrol agents function as a proactive exception detection layer.
An agent that initiates downstream workflows based on defined events or conditions, for example, triggering a reconciliation workflow when a bank statement is received, or initiating an elimination entry when a specific intercompany transaction is posted. Trigger agents are often the first agent in a chain, setting off the specialized agents that carry a workflow through to completion.
An agent that matches transactions across accounts, systems, or entities, applying defined matching logic to identify matches, flag discrepancies, and post correcting entries. Reconciliation agents handle bank reconciliations, subledger-to-GL reconciliations, and account balance verifications.
An agent that automatically generates corrective journal entries to fix discrepancies identified during reconciliation, such as an unmatched transaction flagged by a matching agent or an anomaly surfaced by transaction patrol. Resolution agents complete the loop between detecting an issue and fixing it, producing a properly structured, auditable entry for human review rather than leaving the correction to manual work.
Automation technology that replicates human interactions with software interfaces, such as clicking, copying, and pasting, to execute repeatable tasks. RPA works in stable environments but breaks when data formats or systems change. Unlike agentic AI, RPA cannot interpret context, handle exceptions, or adapt to variation.
A complete, transaction-level copy of the general ledger that Nominal maintains alongside the customer's ERP, including memos, dimensions, and historical patterns. Most finance automation tools work only with summary balances pulled from the ERP; a shadow general ledger gives agents the detail needed to trace a variance back to the specific transactions that caused it.
The oversight layer within an APM platform that assigns agent-generated recommendations to the appropriate reviewer and tracks approval status. Nothing posts to the ERP until a human explicitly signs off, making task management the mechanism that keeps agent execution human-governed rather than autonomous.
The combination of three platform components: a native general ledger, comprehensive task management, and embedded AI agents. Together, let an APM platform execute complete workflows rather than just one piece of them. Competitors typically offer one or two of these components; Nominal's model depends on having all three working together.
The process of pairing individual transactions from two or more data sources, for example, matching a payment to an invoice or an intercompany transaction to its counterpart in another entity's records. Transaction matching at scale is one of the primary use cases for matching agents.
An agent that monitors transactions continuously for anomalies, errors, and compliance exceptions — flagging issues before they reach the close rather than discovering them during reconciliation. Patrol agents function as a proactive exception detection layer.
An agent that initiates downstream workflows based on defined events or conditions, for example, triggering a reconciliation workflow when a bank statement is received, or initiating an elimination entry when a specific intercompany transaction is posted. Trigger agents are often the first agent in a chain, setting off the specialized agents that carry a workflow through to completion.
Now that you know the terms, book a demo to see how Nominal's Agentic Performance Management platform puts them to work in your close.
