Artificial intelligence has recently crossed a critical threshold in the field of financial planning and analysis (FP&A). In the past, AI was used to automate and accelerate workflows that finance teams already had. However, today AI can initiate work autonomously—detecting changes in conditions, launching analyses, and making recommendations on its own—but crucially, human oversight remains in the process.

A recent report from Board calls this change the "Agency Shift" and defines it as the transfer of FP&A initiation from people to advanced systems, while accountability, judgment, and decision ownership remain clearly with humans.

The Agency Shift is happening at a time when companies face critical challenges. Macroeconomic and market pressures have stretched many finance teams thin. Board's research found that high-value work such as decision support and storytelling accounts for less than a third of total FP&A time, while nearly half of working hours are consumed by manual data preparation. Meanwhile, only 4% of organizations can refresh forecasts within a day, and this delay creates a costly time gap between business signals and management responses.

Agentic AI promises to address these challenges, optimizing FP&A in an efficient and scalable way. However, organizations considering deploying these advanced capabilities must establish robust governance structures specifically for AI—AI that not only executes instructions but also makes decisions.

What the Agency Shift really means

Previous-generation AI-assisted FP&A tools could complete tasks assigned by humans—those tedious, time-consuming, and relatively simple processes—thereby freeing finance professionals to spend time on other work. Agentic AI is fundamentally different.

Early tools required analysts to initiate processes, while agentic systems continuously scan for signals and launch analyses as soon as conditions are ripe—whether it's a decline in profit margins, demand fluctuations, or an upcoming cash gap—rather than waiting for the next reporting cycle.

"This is no longer just about automation," says Simone Ferrari, Product Manager for FP&A Solutions at Board. "The real power lies in compressing the distance between events and financial leaders' responses, shortening the interval from fact to decision."

The payoff is not just speed, but also foresight. Through continuous monitoring, these systems can surface emerging issues that teams previously couldn't actually capture, enabling finance teams to respond before signals evolve into crises.

But the same autonomy that narrows the time gap can also widen risk exposure. The more space AI systems have to act autonomously, the higher the potential efficiency gains, but risk exposure increases with autonomy as well. This is why human oversight is crucial.

"AI should be used to support human judgment, not replace it," says Ferrari. "Humans should always keep their hands on the wheel."

Where governance can fail

Governance is often seen as the responsibility of corporate IT or legal departments. But when AI agents are responsible for shaping forecasts or reallocating budgets, accountability falls on finance leadership.

To fulfill this responsibility, finance leaders must shift their mindset toward what Board calls "architected accountability." This means finance departments need to define the rules, thresholds, and boundaries within which systems are allowed to operate, and be prepared to take responsibility for outcomes that are not manually produced.

Developing and establishing these guidelines often serves a diagnostic purpose, exposing governance weaknesses that have long existed but were not previously apparent.

"Agentic AI exposes governance debt that was already there," notes Ferrari. He adds that the first cracks to appear may include: semantic inconsistencies—where different systems encode the same thing in different ways; approval processes defined too loosely to provide clear thresholds for agents; and auditability gaps—where problems emerge when no one can trace where a number came from.

The common thread behind these weaknesses is a lack of explainability—the ability to see and demonstrate how a system arrived at its conclusions. For leaders who must defend AI-assisted decisions to auditors, boards, or regulators, this explainability is essential.

“"The AI did it, I don't know why"—that's not an acceptable answer," notes Ferrari. "Leaders need to know what data was used, which version or scenario the data came from, what drivers shaped the output, and what thresholds were applied."

Ferrari says this explainability—and the visibility that underpins it—is built into Board's platform.

"Our agents are designed with transparency as a principle. For every prompt, users can see the exact data source, the selected data, the company, and even the currency," says Ferrari. "Explainability is engineered in, not added as an afterthought."

Recommendations for getting started

For companies looking to harness the power of the Agency Shift, Ferrari recommends a phased approach, starting with use cases that generate the most value, enabling AI to amplify returns on the highest-impact work.

"Start where AI can make the biggest difference, measure the value it creates, then learn, adjust, and scale," Ferrari advises. "This way, each new use case builds on proven confidence and control."

With this approach, each successful deployment strengthens the governance foundation and paves the way for the next step. When capabilities and governance advance together in this way, organizations will be well-positioned to fully unlock the potential of agentic AI to transform FP&A operations.

Contact Boardto learn more about building infrastructure to support the Agency Shift in FP&A.