From Explanatory Chatbots to Agentic AI Finance
Agentic AI finance refers to intelligent systems that not only analyze financial data and provide insights but also execute well-defined accounting and treasury tasks autonomously under human-approved controls and policies. This marks a shift away from AI chatbots that answer questions without changing the underlying workflow. In finance, that difference matters: chat interfaces help users understand reports, while agents log into systems, follow standard operating procedures, and move transactions forward. Vendors are building ERP-adjacent AI agents that sit alongside core systems rather than replace them, so teams can keep their general ledger as the system of record while delegating high-volume, rules-based work. Finance leaders now ask less about how to experiment with AI and more about when they can trust it with recurring close activities, reconciliations, payments, and liquidity actions that previously consumed staff time.
ERP-Adjacent AI Agents and Autonomous Accounting Work
In ERP and accounting, the frontier is shifting from conversational copilots toward ERP-adjacent AI agents that take on autonomous accounting work with human oversight. As Nominal’s leadership explains, chatbots may summarize or explain a ledger entry, but they do not remove work from the user’s task list. Their platform focuses on agentic performance management that follows standard operating procedures, handles intercompany transactions and reconciliation, and stays ERP-agnostic so it can sit alongside existing systems. These finance automation execution capabilities are aimed at high-volume, multi-entity environments where small efficiency gains compound quickly. The agents prepare entries, run checks, and present proposed actions to accountants, who remain accountable for approvals. This model keeps determinism and traceability at the center: actions are logged, rules are explicit, and the ERP remains the final source of record, while the agentic layer handles the repetitive operational grind.
AI-Driven Treasury Management Becomes Execution Infrastructure
Treasury is emerging as a proving ground for AI-driven treasury management that goes beyond dashboards into orchestrated actions. Kyriba describes treasury platforms moving from system-based visibility to AI-orchestrated liquidity, payments, and risk decisions. Its Trusted Agentic AI is embedded inside workflows for cross-border payments, short-term investing, and daily FX programs, turning the platform into execution infrastructure rather than a passive reporting tool. Through a collaboration with Circle, stablecoin settlement using USDC is brought into the treasury environment, with controls and auditability inside the same system where approvals already live. A second collaboration embeds J.P. Morgan’s Morgan Money, so recommendations on investing surplus cash can move from analysis to execution in one flow. According to Kyriba, customers using its Advanced Liquidity Planning reduce planning time from 10 hours per week to 1.3 hours and improve cash yield by up to USD 2.07 million (approx. RM9.5 million) annually.
Governance, Stablecoins and the Role of Human Oversight
As finance automation execution expands, governance and human oversight remain non-negotiable. Kyriba’s approach to stablecoin settlement shows this clearly: the integration with USDC is framed not as a speculative crypto add-on but as a controlled payment option inside established treasury processes. The GENIUS Act reduced regulatory uncertainty, but adoption still hinges on policy design, audit trails, and executive confidence. To support this, Kyriba and the Association for Financial Professionals are launching a Stablecoins & On-Chain Liquidity in Treasury Certificate, making stablecoin literacy a formal skill for treasury staff. On the AI side, both Kyriba and Nominal stress that agentic AI finance is not ownerless automation. Agentic tools surface recommended actions based on constraints like liquidity needs, FX exposure, and policy rules, while finance teams adjust and approve. This balance keeps CFOs in control as ERP-adjacent AI agents move closer to real-time execution.
What CFOs Should Do Now About Agentic AI Finance
For CFOs, the rise of ERP-adjacent AI agents is less about chasing hype and more about redesigning operating models. The first step is identifying task clusters where autonomous accounting work can be clearly defined: reconciliations, intercompany settlements, daily liquidity sweeps, and hedge execution policies are strong candidates. Next comes integrating agentic platforms that connect to existing ERP and treasury tools without fragmenting the system of record. Vendors like Nominal and Kyriba illustrate two sides of the same shift: one focused on intra-ERP finance operations, the other on AI-driven treasury management and liquidity planning. Governance must evolve alongside technology, with clear approval thresholds, audit logging, and education programs for staff. Finance leaders who embrace AI agents under disciplined controls can reduce manual workloads, respond faster to cash and risk signals, and keep humans focused on exceptions and strategy rather than routine processing.