Agentic AI: how can CFOs prepare for the next era of autonomous finance

On 30 November 2022, people all over the world woke up to learn that a new tool had just been made available to the general public. Nicolas Boucher, who had been working in finance for over a decade at the time, described it as “waking up to Christmas morning when I was a kid. I had a shiny new toy that I couldn’t stop using”. That gift was ChatGPT, and without Nicolas or anyone else knowing it at the time, it would change the way people all over the world do business. And finance is no exception.

What was once a fun platform where one could write poems to loved ones or rap songs to impress friends soon became the new standard of work. During a panel session in Amsterdam in 2024, Boucher recalled how he started exploring the possibilities for finance. “All of my career, I’ve brought finance and technology together to provide value. I saw potential in ChatGPT early on, when almost no one was talking about it for business, let alone for finance”, he shared with the audience.

Today, when asking people in a room how many are using AI for work, about half the room will stand up. And the numbers keep on rising. With the possibility to automate manual tasks such as reporting and analysing expenses, AI has quickly become a part of the financial toolkit.

This new era of AI promises greater inclusion and efficiency, but it also introduces new ethical and regulatory complexity that CFOs need to consider.

But if until recently it was mostly about augmenting people’s capabilities to become more efficient, we are now entering the era of Agentic AI, a new generation of systems that can act independently, make decisions without constant human input, and automate processes even further. But what does this mean for finance leaders?

Gartner’s 2025 Hype Cycle for AI in Finance predicts that agentic AI will have a transformative impact on finance within the next three years, particularly in autonomous forecasting, anomaly detection, and financial control. For CFOs, this means shifting from AI automating tasks that support decisions to it actually making those decisions, and guiding their teams through the changes it brings.

But that’s not all. As AI takes on a more autonomous role, finance leaders must also manage the risks that come with it and decide where the line between machine efficiency and human oversight should sit.

Agentic AI refers to systems that can plan, execute, and adapt actions toward a defined goal using data, context, and feedback loops, rather than explicit step-by-step instructions or prompts. The term “agentic” refers to their capacity for agency, meaning they can act independently and with minimal human intervention.

Agentic AI in finance can perform tasks, but also interpret intent, prioritise outcomes, and learn from its results. It identifies anomalies in real time, initiates corrective actions, manages cash flow fluctuations, and even predicts working capital needs. All while finance professionals focus on their own tasks, without having to revise the AI’s outputs to make sure they’re correct.

The World Economic Forum describes this new era of AI as one that promises greater inclusion and efficiency. It also introduces new ethical and regulatory complexity that CFOs need to consider, as more financial processes and decisions are left to autonomous AI systems.

Traditional automation in finance has focused on rules, with AI executing clearly defined processes such as invoice approvals, reconciliations, and expense audits. Agentic AI, on the other hand, is focused on intent. It interprets goals, learns from feedback, and acts based on that knowledge.

This opens the door to new applications in finance. A Forrester study commissioned by AWS Marketplace highlights the three main areas that agentic AI is transforming: customer service, financial guidance, and financial operations. For the first two, the most common use case is to implement AI agents that interact with customers, helping them manage their accounts, complete loan applications, and monitor their financial situation to identify investment opportunities.

CFOs are still responsible for any decisions made by Agentic AI systems.

When it comes to financial operations, agentic AI systems work behind the scenes to automate complex processes that, so far, have required human expertise. AI was already being used to monitor spending patterns in real time and forecast changes before they occur, but agentic AI can now automatically trigger the necessary corrective actions.

This autonomy brings finance teams a significant amount of time and data, shortening month-end processes and helping them move from reactive to more proactive and predictive functions by:

• Identifying anomalies
• Reallocating budgets
• Reconciling entries

Instead of having to verify transactions and AI outputs, finance leaders can focus on interpreting insights, shaping strategy, and advising the business in real time.

But autonomy doesn’t erase accountability. CFOs are still responsible for any decisions made by Agentic AI systems. This means they now need to monitor not just what decisions are made, but how and why, to ensure they align with the organisation’s mission and policies.

In an article for the Cambridge Judge Business School, Bryan Zheng Zhang and Kieran Garvey, from the Cambridge Centre for Alternative Finance, explain how an “overreliance on AI-driven decision-making without robust oversight could undermine trust, introduce new risks, [and] amplify biases”. In this new environment, CFOs become the ethical gatekeepers of autonomous finance, guaranteeing that every agentic AI system holds up to compliance and ethical standards.

To make this work, finance leaders should set three foundations.

1. Governance: by establishing transparent oversight for every AI-driven process;
2. Auditability: ensuring full traceability of agentic AI decisions and the data that informs them;
3. Ethics: having a clear code of conduct for automation that aligns with company values and regulatory standards.

Despite being in its early stages, agentic AI is already becoming a business priority. According to the Forrester/AWS study, 88% of finance leaders agree they have to innovate faster to remain competitive, which includes moving toward more autonomous financial systems.

Businesses don’t need to rebuild their entire tech stack, but they must ensure strong data discipline and organisational readiness for AI adoption.

However, of the organisations surveyed, 57% are still in the development stage of what they need to take full advantage of agentic AI technology. Establishing these three principles early on helps build internal and external trust in AI, fast-tracking its adoption.

To prepare for the era of autonomous finance, the authors of the Cambridge Judge Business School article reinforce the need to set a strong foundation. They recommend “creating high-performance explainability frameworks” that enable continuous auditability of AI models as they evolve, as well as “algorithmic accountability, and rigorous governance models”.

This doesn’t mean organisations need to rebuild their entire tech stack from scratch, but it does mean ensuring a strong data discipline and organisational readiness for AI adoption.

To make sure they’re ready for agentic finance, CFOs can start by:

• Strengthening data foundations, to ensure data is clean, consistent, and accessible across systems;
• Provide training to equip finance teams with the skills needed to oversee, interpret, and challenge AI decisions;
• Create cross-functional AI taskforces that bring together finance, risk, IT, and compliance teams, and anyone else needed to incorporate agentic AI systems into existing operations;
• Introduce agentic tools gradually through controlled pilots with transparent feedback loops and human oversight.

Implementing Agentic AI is more than a technology project. By approaching it as a governance challenge and following these steps, finance leaders can make adoption faster, safer and more strategic for the whole team.

Agentic AI marks a turning point for finance. What started as process automation is evolving into systems that can think, decide, and act, which fundamentally changes how finance teams work and how organisations move, plan, and grow their financial operations.

This evolution doesn’t mean the end of human oversight in finance, but rather the start of a new phase in which AI systems can operate autonomously and humans hold them to governance and ethical standards.

The next few years will test how ready CFOs are to trust machines with autonomy, and themselves with accountability. They will also prove whether or not Gartner’s predictions were correct. But no matter what the future brings, one thing is certain: those who set clear boundaries for agentic AI today will shape how it behaves tomorrow.