Every CFO is being asked the same question by their board: what are we doing about AI?
Few have a confident answer. Not because the options are lacking, but because there are too many. New tools every week. An AI story from every vendor. A LinkedIn feed that suggests everyone else has automated everything.
So, at our recent AI in Finance Summit, we set out to move past the marketing spin and focus on the facts. To make that happen, we brought together the people building AI, using it and living with it, and asked them to be honest about what actually works.
In the room:
- Duke Nelson, Microsoft, opening the day on why AI transformation is a people challenge before it’s a technology one
- Emma Brown, former CFO at Cube, on leading AI adoption inside a fast-growing, PE-backed business
- Embat, Airwallex and Avalara, on where AI is already adding real value in treasury, payments and tax
- Light and Campfire, on whether finance teams need an AI-native ERP
- Sourav Datta, Google, closing the day on moving AI from pilot to production
Different perspectives. Remarkably consistent message.
The businesses getting real value from AI are treating it as a business transformation – not just another software purchase.
Here are the six insights every CFO should take from the day.
1. It’s Rarely the Tool
Microsoft’s Duke Nelson opened the day with a message that set the tone for everything that followed.
“It’s rarely the tool that’s the problem. It’s everything around it.” – Duke Nelson, Microsoft
Too many businesses buy AI licences and then wait for a return that never arrives. The ones that succeed treat AI as a systemic change. They have a clear business goal, a leadership team saying the same thing, the right governance, and people who are equipped to use the tools well.
That’s what makes AI different from the software that came before it. A new spreadsheet package changed how people worked. AI changes the job itself.
So the skills that matter are shifting. Finance teams need sharper judgement: the confidence to question an output, spot when a number looks wrong and understand where the data came from. And leaders need to think about the pipeline. If AI absorbs the entry-level work that trains junior talent, where do tomorrow’s finance leaders come from?
The good news for CFOs? You’re the natural owner of this. Finance has a view across the whole business, end to end, and sees the full bill.
2. Fix the Foundations First
AI doesn’t fix broken processes. It speeds them up.
Messy data, disconnected systems and siloed teams? AI could amplify those issues instead of solving them. Or, as Avalara put it on the finance operations panel: bad data in, bad data out, just quicker.
Emma Brown, former CFO at Cube, was clear that her AI journey only worked because it followed a digital transformation. The core systems were already in place. AI could then build on them.
“AI amplifies everything. It amplifies the good and the bad.” – Emma Brown, former CFO at Cube
This matters more in finance than almost anywhere else, because every number needs to be right. The tools gaining real traction, like those from Embat, Airwallex and Avalara, combine AI with rules-based controls and governance. That structure is what makes it possible to trust AI with reconciliation, cash forecasting and tax compliance.
But perfect shouldn’t be the enemy of progress. Waiting for a complete data overhaul before you start can mean waiting forever. Fix what matters for one use case, prove it, then build. And get finance and IT working as one team from day one.
3. The Prize Is Scale, Not Cuts
Here’s what nobody said all day: that AI is about cutting the finance team.
For high-growth businesses, the opportunity is scale. Few boards will sign off a finance team that grows at the same rate as revenue.
“No one’s going to approve a linear scaling of finance headcount.” – Theo Wasserberg, Embat
So, the return on AI is measured differently. It’s less about roles removed and more about velocity and capacity: a faster close, a shorter backlog and a finance team that spends less time processing and more time advising.
Emma made flat headcount the explicit goal at Cube, with AI used to keep growing without adding heads. Some businesses are going further still, asking whether a role could be automated before any new hire is approved.
It’s a shift from finance as a back-office cost centre to finance as a driver of growth.
4. AI Has a Running Cost. Budget for It.
With companies’ budget season underway, we put a simple question to the room: who has a dedicated AI budget?
A handful of hands went up. Asked whether that budget covered usage costs, even fewer stayed raised.
That’s a gap worth closing. AI doesn’t behave like a software licence with a fixed annual fee. Google’s Sourav Datta explained why costs are climbing: as businesses move from simple chat tools to AI agents carrying out multi-step tasks, each request uses far more computing power, and the bill grows with it.
Emma shared how she tackled this as a CFO. Her approach was to treat AI spend more like headcount than software:
- Budget by role. A developer and an administrator use AI completely differently.
- Put AI on the org chart. Showing AI agents as “digital employees” helped the board see capacity growing while headcount stayed flat.
- Charge it to the team that uses it, with a separate pot for experiments.
- Set limits, and let people hit them. Nothing makes cost everyone’s problem faster.
Good core systems make all of this easier. When AI spend flows into the ERP and is coded consistently, it can be tracked like any other cost.
5. Start From the Problem, Not the Platform
The AI-native ERP panel, with Light and Campfire, delivered the line of the day.
“We call ourselves the Society for Prevention of Cruelty to Accountants.” – Sam Landsberg, Light
It got a big laugh. But it made a serious point. Finance teams have an extraordinary tolerance for manual pain, and they’ll get the job done regardless. That doesn’t mean they should have to.
The panel’s advice was refreshingly practical. Before you buy any new system, know your baseline: how long the close takes, how many people it involves and how much still lives in spreadsheets. Then find the real bottleneck. A slow month-end is often caused outside finance, by slow approvals or chasing receipts.
Fit matters too. AI-native ERPs suit leaner services and software businesses best today. Businesses with heavy inventory or complex supply chains may find the fit harder, for now.
And whatever you choose, give change management its own line in the budget. Otherwise it quietly disappears.
6. Manage the Hype, Not Just the Technology
For VC- and PE-backed businesses, the pressure to have an AI story is real. Boards see other portfolio companies investing heavily and don’t want to fall behind.
Emma was candid about the risk. Pouring money into AI without a clear objective can create an expensive set of tools and processes that are hard to make profitable later, or to unwind. Her answer was to agree the “why” before the spend. At Cube, every AI initiative had to tie back to one goal: scaling without adding headcount.
She also pushed back on the pressure to label revenue as “AI revenue”. Customers care about outcomes, not the technology behind them. Retention, pricing and customer satisfaction tell the better story.
Often, the gap between what the board wants and what the business needs is one of communication. A roadmap that shows how today’s foundations lead to tomorrow’s AI ambitions can satisfy both.
Google’s Sourav Datta brought it full circle at the end of the day:
“The era of wait and see is almost over.” – Sourav Datta, Google Cloud
The answer isn’t to rush. It’s to start, with purpose.
Looking Ahead
If the day proved anything, it’s that nobody has AI fully figured out. Not Microsoft. Not Google. Not the CFOs in the room. Everyone is learning as they go.
That’s good news. The advantage doesn’t belong to the businesses with the biggest budgets. It belongs to those that build strong foundations, invest in their people, know the outcome they want and move forward one measurable step at a time.
And CFOs are best placed to lead it. Finance sees across the whole business, understands the true cost and can connect AI investment to real value. The finance leaders who lead that change, rather than simply report on it, will shape how their businesses grow.
Over the coming weeks, we’ll go deeper into each of these themes, from building the right foundations and measuring AI’s real return to budgeting for AI and managing investor expectations.
Continue the Conversation
We’re already helping finance leaders turn these insights into action. Right now, we’re:
- Reviewing the finance tech stack of a VC-backed scale-up to find where AI can save time and cost.
- Developing an AI strategy for a multinational, PE-backed organisation.
- Running a cyber and information security audit for a fintech business.
Working through something similar? Get in touch with Cooper Parry’s Tech & High Growth and Cooper Parry Digital teams.
And look out for the next AI in Finance Summit, planned for March.
