AI in Finance and the New Rules of the Game: Lessons from Remote's CFO

The debate in finance is no longer whether to adopt AI. It is how fast, how broadly, and how safely. In CFO Connect's latest live AMA, Michiel Boere, CFO of Remote, spent thirty minutes fielding unscripted questions from the community on AI governance, team adoption, workflow automation, and what it means to run finance at a company operating across over 180 countries.

Drawing on his background as global CFO of UberEats and three years at Remote, Michiel brought operational honesty to a conversation that too often stays at the level of buzzwords.

TL;DR: CFOs should treat AI as a capital allocation decision, not an IT experiment. Michiel Boere argues the biggest risk is underinvesting, while workflow automation and governed enterprise adoption are already delivering measurable productivity gains in finance teams.

What is AI governance in finance?

AI governance in finance refers to the policies, processes, and cultural norms that determine how AI tools are adopted, who is authorised to use them, what data they can access, and how their outputs are reviewed. For CFOs, it sits at the intersection of capital allocation, risk management, and organisational leadership.

Watch the full discussion here:

Key Takeaways

• AI adoption in finance has doubled since 2023, but finance still ranks last among all business functions in actual deployment, according to CFO Connect's State of AI in Finance 2026 report

• The risk of underinvesting in AI is significantly higher than the risk of over-investing, and CFOs need to take a position on that now

• Shadow AI is the new shadow IT; the only two responses when you find an unapproved tool are to shut it down or formally adopt it as an enterprise tool

• Finance team AI adoption is as much a cultural shift as a technical one; demonstration beats mandate

• Real productivity gains in finance begin at stage three of AI adoption: workflow automation, not chat or dashboards

• Cash application and financial statement pre-review are already in production at Remote, delivering measurable time savings

• Start AI automation with processes where errors are recoverable; the general ledger is not where you begin

• Most finance teams will overspend their AI budgets this year; the right frame is organisational tuition, not efficiency ROI

• Token spend is rising exponentially and needs its own line in the budget before the numbers get uncomfortable

It is okay to feel overwhelmed by AI right now. There is so much going on, and it is disrupting every single part of our jobs. -- Michiel Boere, CFO at Remote

Where does the CFO's responsibility for AI begin and end?

The CFO's responsibility for AI starts where capital decisions start: ensuring the organisation is investing at the right level and balancing the risk of moving too slowly against the risk of moving without guardrails.

Michiel opened the session by acknowledging something many finance leaders are reluctant to say out loud: it is okay to feel overwhelmed. His answer to where the CFO sits in the AI governance picture was grounded in a familiar principle: capital allocation. "Ultimately, a CFO is responsible for making good capital decisions in a company," he said. "For essentially making sure that we balance risk and growth and profitability."

He endorsed the view often attributed to Sundar Pichai at Alphabet: the risk of underinvesting in AI is much higher than the risk of overinvesting. The data supports the urgency. Gartner forecasts worldwide AI spending to reach $2.52 trillion in 2026, a 44% year-on-year increase. At Remote, everyone who wants an enterprise AI licence gets one. The approach is permissive but governed, staying flexible as tools shift rapidly from ChatGPT to Claude to whatever comes next.

On shadow AI, his position was equally clear. Active monitoring is in place. When an unapproved tool is found, there are only two options: shut it down or formally adopt it as an enterprise tool. AI policies matter, but culture matters more. When people understand the risk and have good alternatives, the problem largely manages itself.

The risk of underinvesting is much higher than the risk of overinvesting at this moment. -- Michiel Boere, CFO at Remote

What to do this quarter

Audit your current AI investment posture. Are you giving your team enterprise tools, or are they finding their own? A permissive, well-governed access policy is often safer than a restrictive one that drives shadow usage underground.

👉 Watch the full discussion here

Why is finance still behind on AI adoption, and what actually moves the needle?

Finance team AI adoption lags because of culture, not technology. The most effective way to accelerate it is demonstration, not mandates.

The numbers tell a clear story. According to CFO Connect's State of AI in Finance 2026 report, 56% of finance leaders now use AI, double the rate seen in 2023. Yet finance still ranks last among all business functions in deployment, and only 17% are actively using AI in core workflows. A striking 68% of CFOs cite not knowing where to start as the reason for slow adoption.

Michiel's read is that the blocker is cultural, not technical. People need to see what AI can do in their specific context before they try it themselves. His approach is personal: nearly half of his one-on-ones with finance team members are now about what they have automated and how much time it saved. That is the communication strategy. Showing someone that a three-hour task now takes twenty minutes is more persuasive than any policy document.

What to do this quarter

Pick one AI workflow that has already saved time in your team and turn it into a case study at your next team meeting. Concrete examples from inside your own finance function are more persuasive than anything from outside it.

What are the three stages of AI adoption in finance?

AI adoption in finance moves through three distinct stages. The real productivity improvements do not begin until stage three.

Michiel offered a framework for understanding where different teams sit in their AI journey, and it helps finance leaders know exactly where to direct their attention.

Stage 1: Conversational chat. Using tools like ChatGPT or Claude to ask questions, draft text, or pressure-test thinking. This became widely available nearly three years ago and remains many people's entry point.

Stage 2: Dashboards and visualisations. Creating reports, trackers, and data visualisations with AI assistance. Still happening widely across finance teams, but primarily an output improvement rather than a process change.

Stage 3: Workflow automation. AI agents handling recurring, high-volume tasks end-to-end. This is where Michiel sees genuine efficiency gains materialising. Gartner's 2025 AI in Finance Survey confirms the pattern: accounts payable process automation is already the second most common AI use case in finance, with 37% of teams deploying it. McKinsey's 2025 Global AI Survey found that organisations moving AI into production see an average 5.8x ROI within 14 months.

Most finance teams are currently somewhere between stages one and two. The divide between teams that have reached stage three and those that have not is already widening.

The real value, I think, now lies in workflow automation. Things that used to cost someone a lot of time, we can help that person do that much faster. -- Michiel Boere, CFO at Remote

What does production-ready AI in finance actually look like?

The workflows that work are high-volume, rule-based, and recoverable. Cash application and financial statement pre-review are two examples already running in production at Remote.

Michiel shared concrete examples rather than possibilities. The first is cash application: matching thousands of incoming payments to invoices automatically. Complex at scale, time-consuming manually, and crucially, a process where AI errors are recoverable. Just as humans made mistakes here, so does AI, and corrections are straightforward. An agent handles it now, with team oversight.

The second is pre-audit review of financial statements. Before external auditors see Remote's financials, an AI reviews them first. It connects the data, flags anything unusual or missing, and surfaces issues early. What used to require significant accountant time now runs faster and earlier in the process.

His advice on getting AI past internal controls was direct: do not start there. Begin where the cost of an error is low. Build the oversight model. Expand scope over time.

Don't immediately have AI start making entries into your general ledger without any oversight. That is not the place I would start. -- Michiel Boere, CFO at Remote

What to do this quarter

Identify three finance processes that are high-volume, rule-based, and recoverable if something goes wrong. Cash application, document routing, and pre-review checks are all strong starting points that do not require audit sign-off to begin.

How should CFOs budget for AI when the ROI is still unclear?

The honest answer is that most finance teams will overspend their AI budgets this year. The right frame is not efficiency ROI; it is organisational tuition.

Michiel did not soften this. Remote has already exhausted its AI budget ahead of schedule. The Uber CFO reportedly had the same experience in April. This is not a planning failure; it is what happens when usage patterns change faster than budgets can track them. Token spend, the per-query cost of running workflows through AI models, is rising exponentially and in many companies is now rivalling flat subscription costs.

His recommendation: cut elsewhere to fund AI investment, and reframe the overrun as the cost of the organisation learning. Token prices will fall. Cheaper models will become viable. The economics will improve. Getting fluent now is worth the short-term overspend. The underlying logic connects back to his opening point: the risk of underinvesting is greater than the risk of overspending, particularly while the habits and infrastructure are still being built.

For broader context on how global workforce costs are evolving alongside these shifts, Remote's Global Payroll Report 2026 covers pay transparency and retention trends across 6,200 professionals in seven countries.

What to do this quarter

Add token spend as a line item in your next leadership report, even if the figure is rough. The ROI conversation is easier once cost visibility is established, and it becomes easier to forecast once you have a few quarters of trend data.

FAQ: AI, Finance, and the New Rules of the Game

What is the CFO's responsibility for AI governance?

To ensure the organisation is investing in AI at the right level, managing shadow AI actively, and treating adoption as a capital decision rather than an IT question.

What should finance leaders do when they find unapproved AI tools?

There are only two options: shut the tool down or formally adopt it as an enterprise tool. Letting unapproved tools operate in a grey zone is not a safe middle ground.

What are the three stages of AI adoption in finance?

Stage one is conversational chat. Stage two is dashboards and visualisations. Stage three is workflow automation, where real productivity gains begin. According to CFO Connect's State of AI in Finance 2026 report, only 17% of finance teams are currently at stage three.

Which AI workflows are safest to start with in finance?

High-volume, rule-based processes where errors are recoverable: cash application, document routing, and financial pre-review. Avoid writing to the general ledger without oversight until the oversight model is established.

How should CFOs approach AI budget overruns?

Treat them as tuition. The cost of building organisational AI fluency now is worth paying while the technology matures. Token prices will come down and the ROI will improve. The risk of underinvesting today outweighs the risk of overspending.

Where can I learn more about Remote's global payroll and AI capabilities?

Visit remote.com for their employer of record, contractor management, and payroll solutions across 180 countries. Read their Global Payroll Report 2026 for pay transparency data, and their post on Remote's next chapter in global payroll for how they are embedding AI into the platform.

Closing thought: get fluent before the rules settle

The most useful thing Michiel Boere said in this session was also the most uncomfortable: it is okay to feel overwhelmed. The right response is not to wait for clarity.

For CFOs, the practical implication is direct. Give your team enterprise tools, start automating where errors are recoverable, and track where costs are going. Treat the budget overruns as the price of being ready when the economics improve. Sixty-seven percent of finance leaders are already more optimistic about AI than they were a year ago, according to Gartner. The teams driving that optimism are not the ones waiting for a roadmap.

For finance professionals, the takeaway is the same one that came through across every part of this conversation. The teams pulling ahead are finding the workflows that save time, showing their peers, and building from there.

Remote is the intelligent infrastructure for employing and paying people everywhere, helping companies hire, pay, and manage global teams across over 180 countries. CFO Connect is powered by Spendesk, helping finance teams build smarter, more efficient operations with modern spend management tools.