Agentic AI in FP&A: What Finance Leaders Need to Know Now

Artificial intelligence (AI) experimentation is so 2025. This year is all about execution. But how and how fast will your organization adapt?

In our recent webinar, Pras Chatterjee, Global Director of Planning & Analysis and Product Marketing at OneStream, sat down with Jose Melendez, Managing Director at Accenture, and Dhruv Jain, Senior Managing Director at Accenture, to move past theory and focus on what agentic AI means in practice for modern finance teams.

Below are the key takeaways finance leaders should be paying attention to.

Agentic AI Is a Leadership Issue, Not an IT One

Finance has seen no shortage of “transformational” technologies over the years, from ERP to cloud to RPA. Each delivered incremental gains. Agentic AI is different.

As Chatterjee outlined, several forces have converged at once:

Data platforms and cloud infrastructure are finally enterprise‑ready
Large language models can now reason over narrative finance content
AI investments are delivering measurable returns
What was a competitive advantage 12 months ago is rapidly becoming table stakes

Agentic AI doesn’t just respond to prompts; it executes workflows. In financial planning and analysis (FP&A), initiatives like variance analysis, commentary drafting, and scenario evaluation can be handled end-to-end by agents, with humans retaining accountability for decisions.

For finance leaders, agents are a strategic inflection point: AI is shifting from a productivity tool to a digital member of the finance workforce.

GenAI vs. Agentic AI: The Difference That Matters

Melendez offered a clear distinction executives can use internally:

RPA automates tasks
GenAI assists with content and insight
Agentic AI executes toward outcomes

Agentic systems can be assigned goals and operate autonomously across multiple steps while still leveraging familiar technologies like machine learning, rules engines, and analytics. The implication is significant: finance teams can move beyond automating pieces of the process to rethinking the process itself.

What’s Slowing Adoption (and Why It’s Still Fixable)

Despite widespread interest, most finance organizations still rely heavily on manual work. The barriers aren’t surprising — but they are now unavoidable:

Trust, governance, and accountability
Data quality and fragmentation
Legacy infrastructure
Skills and talent constraints
Speed of enterprise change

Data, as Melendez emphasized, remains the foundation. Finance does not own most of the data it depends on, which makes agentic AI adoption an enterprise issue, not just a departmental one.

Closing the Gap Between Perception and Reality

Jain highlighted a common challenge for executive teams: leadership often sees output without seeing the effort required to produce it. Reports arrive on time, but only because teams are manually stitching together data behind the scenes.

The recommendation wasn’t to fix everything at once. Instead:

Start where business value is highest
Focus on one or two high‑impact processes
Build credibility and momentum through visible wins

This approach allows finance leaders to improve data maturity in context, without stalling progress.

Where Finance Teams Are Seeing Measurable Value Today

This conversation focused on results, not roadmaps. Three use cases stood out:

FP&A narrative and commentary drafting
Agents produce first‑pass variance explanations and management commentary, allowing finance teams to focus on judgment, analysis, and business partnership.
Conversational analytics for leadership
Embedding agents alongside dashboards enables executives to ask follow‑up questions directly, reducing back‑and‑forth and accelerating insight.
Scaled variance analysis across complex organizations
In multi‑entity environments, agents tailor insights for business units, categories, and executives, dramatically compressing close and reporting timelines.

Across all three, the takeaway was consistent: agentic AI reallocates time from manual effort to higher‑value decision support.

Human in the Lead: Governance Is Non‑Negotiable

Agentic AI doesn’t eliminate accountability; in fact, it demands clearer ownership.

Jain reframed the idea of “human in the loop” as human in the lead. If a finance leader is accountable for a forecast, scenario, or external narrative, they must approve it. Agents prepare. Humans decide.

Trust, the panel noted, is built the same way it is with new employees: through oversight, metrics, and feedback. The difference is scale, which makes governance, auditability, and monitoring essential from day one.