Finance departments have spent years seeking ways to reduce the administrative work that occupies much of the accounting calendar. Reconciliations, invoice coding, approvals, expense management, data entry, and document collection remain necessary responsibilities, yet many organizations still rely on manual processes to complete them.
AI agents introduce another option for addressing this workload. During the Controllers Council and BILL webinar, “From Transactions to Strategy: How AI Agents Help Reshape Finance,” Philip Peck, VP Finance Transformation at Peloton Consulting Group, joined Jessica Hale of BILL to discuss where automation and AI agents fit within finance, how leaders can determine which processes are suitable for each, and what these technologies may mean for the finance profession.
Why Manual Work Remains So Common in Finance
An audience poll at the beginning of the webinar illustrated how much manual work remains within finance departments. Fifty-nine percent of respondents said their teams manually reconcile ledger entries, while 55% manually categorize transactions and code invoices, and another 55% manually approve and schedule payments. Forty-one percent reported manually collecting forms from independent contractors.
Peck grouped the reasons for this administrative workload into four broad categories: processes, technology, data, and organizational culture.
Many accounting processes developed gradually and were never reconsidered as the organization grew. Spreadsheets, paper receipts, repetitive data entry, multiple handoffs, and complicated approval structures can persist simply because they became part of the established workflow.
Technology presents a related difficulty. An organization may have capable ERP, payroll, expense management, and banking systems, but those systems do not necessarily communicate effectively with one another. Finance employees are consequently left to reconcile information and move data between applications manually.
Data quality and compliance requirements add another layer. Finance departments must maintain accurate records, document transactions, satisfy audit requirements, and resolve inconsistent or incomplete information. Peck also noted that cultural considerations can slow modernization because accounting and finance organizations appropriately place considerable importance on accuracy and compliance.
Start by Understanding the Process
Before deciding what to automate, finance leaders need a detailed understanding of how work is currently performed.
Peck recommended mapping existing processes from beginning to end. Finance teams should document who performs each task, where information originates, which systems are involved, how frequently individual steps occur, and whether every step remains necessary.
Once the process has been documented, leaders can examine several practical considerations. High-volume and repetitive work tends to offer stronger opportunities for automation. Structured and rules-based activities are generally easier to automate than processes dominated by exceptions. Processes with frequent manual errors or disproportionate time requirements may also deserve early attention.
Data quality remains central to that assessment. As Peck explained, “Automation thrives with digital, accessible, high-quality data that has the appropriate lineage, everything around that data.”
Accounts payable offers several readily identifiable examples. Invoice capture and data entry can benefit from optical character recognition and document processing. Purchase order and invoice matching can use rules-based logic. Approval workflows can automate routing and reminders, while payment scheduling can incorporate automated payment runs and exception handling.
Expense management presents similar possibilities, including receipt capture, policy compliance checks, digital approvals, reconciliation, and reimbursement workflows.
Prioritize Work That Can Produce Measurable Results
Finding processes that can be automated does not mean they should all be addressed simultaneously.
Peck recommended considering both potential return and implementation effort. Processes that can deliver substantial benefits with relatively modest implementation requirements are sensible early candidates. Higher-impact projects that require greater organizational effort may be better suited to pilots and phased implementation.
Early projects can also help finance leaders demonstrate measurable progress and establish confidence among employees and other stakeholders.
“Automation success is as much about change management as it is about technology,” Peck said.
He recommended measuring results through indicators such as cycle time, exception rates, and reductions in manual hours. Finance leaders should also continue reviewing workflows after implementation because processes, policies, and business requirements will change.
What Is an AI Agent?
Traditional automation and AI agents address different categories of work, making the distinction between them important.
Peck described AI agents as “autonomous or semi-autonomous systems designed to perceive the environment, make decisions, and take action to achieve specific goals.”
Unlike conventional software that follows fixed instructions, AI agents can use context, learn from previous interactions or outcomes, adjust their behavior, and collaborate with people or other agents.
Their work generally follows an observe, reason, act, and learn cycle. An agent can collect information from documents, applications, data sources, or user input; interpret that information; determine an appropriate action; execute that action; and, depending on the system, use the result to inform subsequent activity.
Within finance, possible applications include reconciling transactions, forecasting cash flow, identifying anomalies or potential fraud, managing invoice approvals, and supporting payment processes. An AI agent could, for example, receive an invoice, extract its information, compare it with a purchase order in an ERP system, identify an exception that requires human attention, and post an accurate entry when appropriate.
Automation and AI Agents Serve Different Purposes
One of the webinar’s more useful distinctions concerned when finance teams should rely on conventional automation and when AI agents may be more appropriate.
Traditional automation is well suited to work that is structured, repetitive, rules-based, high-volume, and predictable. Invoice coding, purchase order matching, expense submission workflows, bank reconciliations, and scheduled report generation are examples.
AI agents become more applicable when the work involves judgment, context, changing circumstances, varied data, or coordination among several systems.
As Peck explained, “Automation focuses on repeatability, execution at scale, AI agents much more around adaptability, judgment at scale.”
Potential agent-based applications include detecting fraudulent or duplicate expense claims, predicting payment delays, prioritizing vendor payments according to cash requirements, classifying transactions through pattern recognition, and drafting variance explanations for management reports.
Finance leaders should consider business impact, risk, data readiness, process maturity, stakeholder complexity, and scalability before determining where a particular technology belongs. Strong data remains especially important because both automation and AI depend upon accessible, reliable information.
A Practical Progression for Finance Teams
Rather than attempting to introduce advanced AI capabilities throughout finance at once, Peck outlined a gradual progression.
The first stage centers on establishing an automation foundation. Organizations can standardize processes, digitize information, improve data quality, and remove unnecessary manual work. Invoice capture and approval routing are examples of activities that can fit this stage.
The next stage introduces AI-supported capabilities such as pattern recognition, expense categorization, and anomaly detection.
More advanced applications involve adaptive AI agents capable of making and executing certain financial decisions while maintaining appropriate human involvement.
Governance must accompany this progression. Finance leaders need clearly documented accountability for AI and automation performance, regulatory and audit transparency, measurable performance indicators, and defined expectations for human oversight. Employees also need training that allows them to supervise and work effectively with these systems.
Giving Finance More Capacity for Strategic Work
The larger opportunity extends beyond reducing administrative hours.
When routine processes require less manual attention, finance professionals can devote additional time to analysis, planning, forecasting, and consultation with business leaders. Agents can also operate continuously, monitor transactions as they occur, identify anomalies, and provide more current financial information.
That can shorten the interval between an event occurring and finance being able to interpret its implications. Instead of devoting substantial effort to assembling information about past performance, teams can spend more time evaluating what current information means for future decisions.
“Finance shifts from reporting on the past to shaping the future,” Peck said.
This change can affect the work of individual finance professionals as well. Analysts who previously spent considerable time compiling expense information, for example, may instead evaluate supplier payment strategies and their effects on cash cycles. Finance teams can devote more attention to scenario modeling, investment planning, forecasting, business performance, and discussions with operating leaders.
The Skills Finance Professionals Will Need
The webinar concluded with an audience question about the skills accounting and finance professionals will need as AI agents become more common.
Peck identified AI prompting and communication, critical thinking, professional skepticism, data literacy, governance knowledge, automation design, exception management, business partnering, AI-assisted financial analysis, technology fluency, and change management.
Professional judgment remains particularly important. Finance employees must be able to examine AI-generated results and ask whether an output is reasonable, complete, consistent with company policy, and based on appropriate assumptions.
As routine activities become more automated, employees may also spend a larger portion of their time investigating exceptions and circumstances that do not fit predefined patterns. That places additional importance on analytical ability, business knowledge, and judgment.
Moving From Transactions Toward Strategy
For finance leaders, the practical starting point is not simply finding places to introduce AI. It is understanding existing processes, deciding which work is suitable for conventional automation, determining where AI agents can contribute additional judgment or adaptability, and establishing the governance necessary to use both responsibly.
Peck summarized the broader objective near the conclusion of the webinar: “AI agents do not replace finance. They amplify the strategic influence of finance across the business.”
That distinction is consequential. Reducing manual work has value on its own, but the larger benefit comes when finance professionals can redirect their attention toward analysis, planning, business consultation, and decisions that influence company performance.
To hear the complete discussion and learn more about identifying automation opportunities, evaluating AI agent use cases, establishing governance, and preparing finance teams for this change, watch the full webinar.
ABOUT OUR SPONSOR:
BILL is the intelligent finance platform trusted by nearly half a million businesses and their accountants to manage, move, and maximize their money. BILL powers businesses ranging from fast-moving startups to growing companies with complex operations. We use AI to deliver strategic finance capabilities in one integrated platform that includes AP, AR, expenses, forecasting, procurement and more. With a member network of more than 8 million, BILL’s platform processes ~1% of US GDP annually. Headquartered in San Jose, California, BILL is a trusted partner of leading U.S. financial institutions, accounting firms, and software providers. For more information, visit bill.com.
