I speak with finance leaders every day and, when it comes to AI, one thing stands out: CFOs are done experimenting. Now they are demanding that AI prove its worth, and many tools are not ready for that conversation.

Recent survey data from Silicon Valley Bank found that CFOs at VC-backed companies expect to spend a median of $50,000 this year on AI tools and platforms, more than double the prior year’s level. The same research suggests AI adoption has become one of the top issues on finance leaders’ minds. That should not be surprising. CFOs are under pressure to improve productivity, preserve capital, support growth, and help their organizations make faster decisions with greater confidence.

Kyriba’s own 2026 CFO Survey tells a similar story. In a survey of 1,400 global finance leaders, 91.9% of CFOs said they are already integrating AI into financial decision-making, either across some processes or virtually all of them. The experimentation phase is giving way to the accountability phase.

But as AI budgets grow, so does the scrutiny attached to them. CFOs are no longer asking whether they should spend on AI. They are asking where AI can create measurable value. For a roughly $50,000 annual AI budget, finance leaders will expect practical outcomes: faster reporting, leaner finance operations, better forecasting, stronger risk visibility, and fewer manual processes.

The winners will not be generic AI tools disconnected from the CFO’s systems, controls, and decision-making responsibilities. The winners will be AI capabilities that are connected, integrated, and embedded within trusted financial workflows.

AI spending has entered the CFO accountability zone

In many organizations, AI began as decentralized experiments. Teams tried copilots, generative AI tools, and productivity assistants to see what worked.

Now that the trial phase is over, the questions CFOs ask have changed. It's no longer about whether a tool can produce impressive outputs—it's about which business process it improves. It's not enough to claim employees will save time; CFOs want to see value in the income statement: savings, avoided future spend (including headcount avoidance), and increases to interest income through more optimal liquidity decisions. If the AI does not prove its value, it won't survive the next budget cycle.

The CFO’s AI scorecard is different

For finance leaders, AI value is not measured only by adoption rates or user enthusiasm. It is measured by outcomes.

CFOs want to know: Will the forecast improve? Can treasury invest earlier and for longer? Can we answer board questions quicker and more effectively? Can we drive more insight and data-driven decisions without adding headcount?

In treasury, that scorecard becomes very practical. If AI can help a team identify idle cash sooner, explain a cash forecast variance faster, detect a payment anomaly earlier, or proactively deliver risk exposure and liquidity scenarios that highlight opportunities to protect cash, the value is no longer theoretical. These are the kinds of use cases Kyriba is focused on: applying intelligence to liquidity, payments, forecasting, and risk workflows where faster insight can translate into better financial outcomes.

These are the questions that matter.

A $50,000 AI budget may not sound transformational in isolation, particularly for larger enterprises. But for many growth-stage organizations, that level of spend can be meaningful. It often represents a test of whether AI can earn a larger role in the operating model. If the investment delivers measurable value, CFOs will expand it. If it produces scattered experiments and unclear benefits, budgets will tighten quickly.

Here is what $50,000 buys in traditional finance terms: roughly half an FTE analyst, a mid-tier BI tool subscription, or a few weeks of advisory support. CFOs evaluating AI are not comparing it only to other software. They are comparing it to hiring, outsourcing, or living with the status quo. That is the bar AI has to clear.

A reasonable CFO expectation? A 50% time reduction in time spent, avoidance of one incremental hire or an AI-driven liquidity plan that reduces borrowing and interest expense. That is measurable. That is defensible. That is the kind of ROI that survives budget scrutiny.

The most valuable AI use cases are close to the work

One of the biggest mistakes organizations can make is treating AI as a layer that sits apart from the workflows where decisions actually happen.

For CFOs, value is created inside financial processes: reporting, forecasting, cash management, liquidity planning, risk management, payments, compliance, and performance analysis. AI disconnected from those workflows may be useful, but it will not become indispensable.

Consider fraud prevention. In a recent conversation on Liquid: How CFOs Outperform, Sift CEO Marc Friend described how fraud prevention teams use machine learning to distinguish good actors from bad actors, improve acceptance rates, and manage risk without turning every transaction into a roadblock. That is a useful model for CFOs evaluating AI more broadly: the value is not the algorithm by itself. The value is the decision it improves, the friction it removes, and the risk it helps control.

The same principle applies to treasury and finance. AI is more valuable when it is connected to the systems, data, and controls teams already rely on to manage cash, payments, liquidity, and risk. A general-purpose AI tool may help someone draft a summary. But when intelligence is connected to bank data, payment workflows, forecast inputs, risk exposures, and approval processes, it can help finance teams act with more speed and confidence.

The strongest use cases share three characteristics: they reduce manual effort in recurring finance activities, they improve decision quality by making change and risk more visible, and they improve trust and control in decision making.

Too many AI tools promise to 'transform finance' but deliver glorified summarization. A chatbot that can draft an email or answer a question about last quarter's revenue is useful, but it's not measurable. CFOs need AI that can forecast cash and plan liquidity.

AI should help finance teams run leaner and smarter

CFOs expect AI to reduce the number of new hires needed to support greater complexity and quantity of work. Perhaps even reduce the size of their teams, especially where staff are dedicated to manual tasks. Within our own customer community, we are already observing a deceleration in the addition of new platform users. The data suggests organizations are waiting on new hires, which is all the more reason for AI to prove itself now. CFOs are waiting. Their teams are waiting. Value-added projects are waiting.

And this is the important point. While there is indeed curiosity about whether finance teams will shrink because AI potentially takes away those jobs, the more realistic expectation is that AI will empower teams to achieve more. The same people, especially the high performers, deliver at a higher level. This is where the value of AI will be measured. Headcount avoidance may be part of the savings CFOs can expect. But headcount reduction is much less likely than teams like treasury, FP&A, A/P and A/R empowered with AI delivering measurable value that impacts the bottom line of the income statement.

Trust will separate durable AI from disposable AI

CFOs have good reason to be cautious. Financial workflows carry consequences. Poor data, weak controls, or inaccurate outputs can create real risk.

That is why trust will become one of the defining criteria for AI investment. CFOs need to trust the outputs, trust the recommendations and ultimately know they can rely on AI-driven decisions. Wherever human-in-the-loop is placed, ultimately AI needs to “show its work” like a 9-year-old solving a math problem. You know it’s right when you see the documented proof. Reasoning matters.

Trust in AI outputs is also earned by having clean data which the CFO’s teams can similarly trust. If you trust the ingredients, there is a far greater likelihood the recipe will meet expectations.

For CFOs, the takeaway is simple: do not evaluate AI only by the quality of the answer. Evaluate the data behind it, the controls around it, and the workflow it supports. This is where Kyriba’s role matters. Finance teams need AI that works inside a trusted environment for liquidity, payments, forecasting, and risk, with the governance and visibility CFOs require.

The next phase of AI value belongs to the CFO

The CFOs who get the most from AI will not be the ones chasing the latest feature. They will be the ones who start with a real business problem, such as cash forecasting, liquidity planning, payment risk, or more efficient currency hedging, and ask, “How can AI make this faster, cheaper, or better?” That is the mindset that turns a $50,000 experiment into a $1,000,000 win.

CFOs are well suited to own that process because they can connect AI investment to financial outcomes, operational discipline, and enterprise risk. For CFOs, the mandate is not to chase AI. It is to make AI accountable.

So here's the test: pick one task that's manual, slow, or error-prone. Measure how long it takes today. Then ask how AI can improve that process - and subsequently quantify the impact of that improvement in balance sheet, income statement or cash flow terms. If you like the result, you've found a use case worth funding. If the answer is no, keep looking. AI spending in 2026 won't be judged on potential. It'll be judged on proof.