Last week, we laid out the numbers: 59% of finance functions claim to be "using AI," but only 11% have meaningfully implemented it. The reader response confirmed what Larry and I suspected. The gap is real and expensive, and most finance teams are stuck somewhere in the middle.

Several of you shared your own versions of the same story. AI tools are scattered across departments with no clear ownership. Pilots that started with enthusiasm and stalled without structure. One reader put it simply: "We have activity, but no capability."

That's the diagnosis. This week, Larry and I are introducing the solution.

A Finance AI Center of Excellence is not what most people think it is. It's not a data science team, innovation lab, or another committee. It is a structural mechanism designed to convert scattered AI activity into governed, scalable capability that finance can actually rely on.

Larry will walk through the framework and provide important guardrails to ensure successful deployment. I'll add the practitioner perspective: what this looks like in practice, where it breaks down, and what most companies get wrong.

Next week, in the final edition of this series, we'll give you the implementation roadmap: how to actually build your CoE, even with a small team.

An AI Center of Excellence is often misunderstood. Most people hear "CoE" and picture a centralized data science team or an IT-led innovation lab. For the Office of the CFO, that model misses the point entirely.

A Finance AI CoE is about structural alignment. Its purpose is to convert scattered AI activity into a governed, scalable capability aligned with finance's core responsibilities: accuracy, control, insight, and trust. It is not about experimentation. It is about turning experiments into reliable operations.

Larry identifies four value propositions that make a Finance AI CoE worth building. Each one addresses a specific failure pattern that shows up when finance teams scale AI without structure.

Unlike other functions, finance cannot afford "black box" AI.  Models must be explainable. Output must be auditable. Controls must be embedded, not bolted on after the fact.

A Finance AI CoE establishes governance standards: model validation and explainability requirements, alignment with internal controls and external audit, and clear ownership of finance AI assets. This directly addresses the top concern cited in both the Gartner and L.E.K. research: managing risk.

Anna's take: Every company needs an AI policy. Finance needs it several times over. Without a policy, you get two extremes. Some teams freeze and do nothing, afraid of making a mistake. Others act recklessly, moving fast with sensitive data and no guardrails. It can be two pages or twenty pages, but any policy is better than no policy. And a policy you revisit quarterly is worth ten times more than one you wrote once and filed away.

Many CFOs prefer AI embedded within existing finance platforms (ERP, EPM, close systems) rather than standalone point solutions. A CoE enforces this architectural discipline. It prevents tool sprawl, reduces integration risk, and ensures AI enhances existing workflows rather than creating parallel, redundant processes.

The CoE is responsible for ensuring that the future-state operating model for finance can be achieved using tools embedded and integrated into the finance ecosystem, not bolted on from the outside.

Anna's take: There is the tension nobody talks about. General-purpose LLMs are progressing much faster than the AI embedded in your existing software. Embedded tools are better for security and data integrity, but they are often delayed and less capable than what you can build yourself. Meanwhile, new tools launch weekly, vendor claims are hard to verify, and "AI-powered" has become a marketing label more than a capability description. Navigating this space is noisy, confusing, and time-consuming. That is exactly why you need a CoE: someone whose job it is to separate clear signals from noise and make the architectural calls.

Often,  AI pilots linger because success is poorly defined. Finance AI CoE reframes AI initiatives in CFO language:

By tying AI use cases directly to finance KPIs, the CoE turns innovation into performance management.

Anna's take: Let’s keep in mind that not every AI initiative shows a direct positive ROI, and that is fine. There is a difference between AI enablement and AI investment. Some of the real value shows up where finance does not traditionally look: no overtime on close weekends, higher retention, better team morale. Those savings often sit in the HR budget, not the finance P&L. If you only measure the obvious metrics, you will kill initiatives that are actually working. A CoE ensures you are tracking the full picture, not just the spreadsheet-friendly part.

Few studies suggest that AI is eliminating finance jobs en masse. Instead, most indicate that roles are evolving.

Finance AI CoE helps manage this transition by:

Without this structure, AI adoption remains superficial and fragile.

Anna's take: Larry is right, and it goes deeper than role redesign. The bigger structural shift is this: we will be working in and managing mixed teams of people and AI agents. That is not a theoretical future; it is already happening. The managerial skill set must evolve to keep pace. Distributing tasks across a mixed team, knowing what to delegate to AI versus to a person, and reviewing AI output with the same rigor are new competencies that most finance leaders have not yet developed and need to do to ensure sustainable success and competitive advantage.

The most important insight from both the Gartner and L.E.K. research is that finance functions are using AI, but they are not yet AI-enabled.

Being AI-enabled requires that AI is embedded in core finance processes. That governance and controls are designed in, not bolted on. That value is tracked and optimized continuously. That information is trusted enough to act upon.

That leap does not happen organically. It requires intentional leadership and a formal mechanism to scale responsibly.

Without a formal CoE, finance organizations tend to repeat the same failure patterns:

And here is the uncomfortable truth Larry highlights: many of these risks increase as AI adoption grows. More tools, more models, and more data without governance create fragility, not advantage.

Before you read on, take 3 minutes to find out where you stand. I built an interactive Red Flags Diagnostic that scores your organization across all four value propositions. It is free, anonymous, and will tell you exactly which areas need attention first.

Now you know the framework and where your gaps are.

In the subscriber section, three stories from my practice will help you understand what to look for: what actually goes wrong when finance teams scale AI without structure, and how to spot the same patterns in your organization before they become expensive.

Two down, one to go.

We started this series with the uncomfortable truth: most finance teams are using AI but not in a way that scales, governs, or creates lasting value. This week, Larry and I laid out the structural solution and what happens when it is missing.

Next Tuesday, we bring it all together. Larry and I will walk through the implementation roadmap: how to actually build a Finance AI Center of Excellence, even with a small team. Role definitions. 90-day milestones. How to position this to your C-suite without it sounding like another initiative that dies in committee.

See you next Tuesday.

Anna

This newsletter is for you, and we want to make it as valuable as possible. Please reply to this email with your questions, comments, or topics you'd like to see covered in future issues. Your input shapes our content!

Want to dive deeper into balanced AI adoption for your finance team? Or do you want to hire an AI-powered CFO? Book a consultation!

Did you find this newsletter helpful? Forward it to a colleague who might benefit!

Until next Tuesday, keep balancing!

Anna Tiomina
AI-Powered CFO