AI leadership for the CFO

AI leadership for the CFO means actively steering AI adoption within the finance function — not leaving the technology to the market or to an enthusiast in the team, but setting direction, prioritising, and making sure the team comes along. Four building blocks under an AI-enabled finance function: vision and purpose, governance and guardrails, people and skills, and platform choices — the CFO is the natural sponsor across all four.

Technology isn't the limiting factor in AI adoption in finance. Copilot, Claude, and accounting integrations are available, good enough, and affordable. What's missing in most SME and scale-up finance functions is leadership: a CFO who sets direction, prioritizes, and makes sure the team comes along. Leaving AI in finance to "the market" or to "an enthusiast in the team" produces fragmentation, shadow AI, and ultimately disappointment. Actively steering it produces, within a year, measurable results in close time, cash position, and audit quality.

For the CFO specifically this is a natural role: AI touches finance directly (tooling budget, productivity effect on personnel costs, possible redefinition of service-delivery margin), and the CFO sits in the boardroom where these decisions are made.

The four building blocks of an AI-enabled finance function

A finance function successfully integrating AI rests on four building blocks. Remove one and the rest collapses.

1. Vision with urgency from the CFO

Without a CFO who uses AI themselves and visibly carries the message that this is the direction, nothing structural happens in the team. Individual controllers experiment, one AP clerk builds something nice, but the finance function as a whole does not change. The explanation is simple: AI adoption requires people to do their work differently. They don't, without explicit direction and air cover.

The most effective CFOs voice public commitment, use the tools themselves in their daily practice (analyses, board prep, mail), and are willing to free up time and budget for it. That's not a talk at a kickoff. That's a reorientation of what's valued inside the finance function.

2. Business-first strategy

AI adoption in finance doesn't start with "which tool do we buy?" but with "which finance processes do we want radically different a year from now?" Close from 8 days to 5. DSO from 50 to 40. VAT reconciliation in half the time. A forecast that holds up every morning. The technology is the means, not the goal.

Finance teams starting with a tool — "we have Copilot, now we have to do something with it" — produce PowerPoint exercises. Finance teams starting with a business goal — "variance commentary has to go from 4 hours to 1" — produce results.

The CFO's role is to make this reversal real. Every AI investment must be tied to a concrete finance process, with a concrete owner, and a concrete metric to say in three months whether it has worked.

3. AI literacy across the entire finance team

AI literacy isn't optional since February 2025 (see EU AI Act for finance teams). But the AI Act obligation is the minimum. The real goal is for finance staff to experience AI as part of their normal toolbox — not as something scary, not as something "IT" does.

That requires hands-on training, not e-learnings. An afternoon where a controller sits next to an experienced AI user and together they make a real variance commentary with AI support is more effective than ten hours of video. For a finance team of 12, that means an investment of roughly 40-80 training hours per year per person in the first year, dropping to a fraction in later years.

4. Agile, pilot-based execution

Large finance AI programs supposed to deliver in two years don't work. What does: a list of 10-20 possible finance use cases, prioritization on impact-feasibility, then two or three per quarter run as pilots. What works scales, what doesn't gets called off.

The quadrant: impact (hours, cash, audit quality) on one axis, feasibility (data, integration, team capacity) on the other. Low-hanging fruit — high impact, high feasibility — first. Gold nuggets (high impact, low feasibility) reserved for later. See Scoring use cases for finance.

The finance AI champion as pivot

An AI champion in finance isn't a job title — it's a role someone takes on. Traits: technically literate enough on AI to evaluate tools, finance-literate enough to judge use cases, with enough standing on the team to push things through.

In a small finance team (< 8 FTE), the finance manager or the CFO is often the champion themselves. In a larger finance team (12-30 FTE), it's typically a senior controller, a lead AP/AR, or the finance manager. The important thing is that this person:

• Is the contact point for questions on which tool is allowed for what.
• Maintains and updates the AI policy for finance.
• Makes successes inside the team visible so others join in.
• Bridges finance, IT, and compliance — not on behalf of one of them, but above those silos.

One champion per 10-15 finance FTE is a workable ratio. Larger finance functions have a network of champions per sub-domain (AP, AR, close, reporting).

Proactive vs reactive leadership

There are two CFO postures around AI. The reactive one waits on IT, on legal, on a vendor pitch. "We'll look when we have to." Result: the competitor or sister company is first, employees use shadow AI in the meantime, and the eventual adoption is defensive and incomplete.

The proactive CFO flips this. He or she writes a first version of the finance AI policy themselves, presents it to IT and legal, appoints a champion, and frees up budget for training and tools before it's needed. This isn't recklessness — it's steering. You make the risk a topic of conversation instead of waiting it out.

The 90/10 principle for finance budget

A practical observation: successful AI adoption in finance is 90% people work and 10% technology. That means time and budget spent on changing behavior, training the team, designing governance, and celebrating successes deliver a higher ROI than time and budget spent on the latest tool.

Practically: if you have €100K for finance AI in a year, spend €10-20K on licenses and tooling, and €80-90K on training hours, pilot coaching, and internal communication. Not the other way around. For CFOs accustomed to treating tools and software as concrete cost items and training as "overhead," this is counter-intuitive — but empirically what works.

Where this lands in practice

• An 80-FTE family business makes the CFO responsible for finance AI adoption. He writes the AI policy, rolls out Claude Team and Copilot, organizes a lunch session every two weeks where a controller demonstrates a concrete use case. After six months 80% of finance staff use AI daily for real work.
• A 25-person consultancy lets the finance manager spend one hour per week on AI experiments and share his learnings in the management meeting. Within a quarter every senior colleague has automated at least one recurring finance task.
• A 150-FTE manufacturer appoints an AI champion per finance sub-domain (AP, AR, close, reporting — 4 people) who catch up with the CFO monthly. Use cases come up bottom-up, get prioritized top-down. In year one the company frees up 2 FTE of finance productivity, immediately redeployed into a richer FP&A function.
• A skeptical CFO concedes after three months. He now uses Claude daily for his board briefings, Copilot for summaries of Teams meetings, and a Saldus Q&A agent for direct questions to the ledger. His calendar has freed up half a day per week for strategic work. Inside finance, this is the most convincing statement.

The first quarter — concrete

For a CFO wanting to start this quarter:

• Weeks 1-2: pick two AI tools, sign enterprise contracts, appoint a finance AI champion with half a day per week.
• Weeks 3-4: write a two-page AI policy for finance (tier classification, approved tools, basic rules for MCP and integrations), wrap up with legal and IT, communicate to the team.
• Month 2: train the entire finance leadership (controllers, finance manager, CFO themselves) in one day on capabilities, prompts, context, and governance. Every team lead names one use case from their own work.
• Month 3: start two pilots with concrete finance KPIs (close time, DSO, variance-commentary time). Weekly standup of 15 minutes on progress and obstacles.

After one quarter you have AI-capable finance leadership, a working policy, and the first measurable results. That's enough foundation to build on. Waiting for more certainty before starting is itself the most expensive choice.