Glenn Hopper spent years on both sides of the table: as a CFO accountable for ROI and as a consultant brought in when things don’t work. That combination gives him a very specific POV on AI investment in finance.

His take: finance leaders are at risk of investing in the wrong things in the pursuit of ROI. The most important thing for them to rubberstamp, even if it gets them no hard gains, is the fundamentals.

That’s because in finance – where decisions need to be explainable, auditable, and defensible – your messy data, undocumented processes, and unclear accountability won’t fly in the AI era. Here’s where you should invest to do this right in 2026.

1. Invest in data alignment across systems

This is your top priority.

Years ago, finance could rely on the general ledger. Today, meaningful financial insight depends on data from ERPs, CRMs, billing systems, project tools, AP/AR platforms, and often multiple versions of each.

“Sometimes companies have multiple ERPs. Sometimes forty of them,” Glenn says.

AI can’t reason across that chaos unless leaders do the unglamorous work of alignment first.

That investment looks like:

• Identifying a single source of truth when systems disagree. If a billing contact or number exists in both the CRM and the ERP but doesn’t match, leaders have to decide which one is the source of truth – otherwise AI won’t know what to trust.
  
  
• Standardizing metric definitions so everyone is measuring the same thing. Even basic financial metrics, like Days Sales Outstanding (DSO), are often defined differently across teams. If people aren’t aligned on how a metric is defined, AI can’t automate or analyze it correctly because it has no consistent rule to follow.
  
  
• Making sure AI can technically access the right systems via APIs. Even in large enterprises, systems that should be integrated often aren’t. For AI to work, it has to be able to point at and pull from all the relevant systems, which requires basic technical setup like APIs so the data can actually flow where it needs to.

This doesn’t always require building a massive data lake, but it does require investment that most finance leaders have avoided because the ROI feels abstract.

2. Invest in finance-specific AI literacy, not blanket training

For finance teams, the issue with AI adoption isn’t curiosity, it’s risk. Without clear training and guidance, people don’t know:

• What proprietary or PII data they’re allowed to use
• Which environments are secure versus unsafe
• Whether AI outputs are acceptable in audit-sensitive workflows

So finance teams need to understand:

• How LLMs actually generate outputs, so results aren’t treated as authoritative just because they sound confident. If finance leaders are going to offload decisions to AI, they need to look under the hood a little bit. Otherwise, using AI is no different than asking a Magic 8 Ball.
  
  
• Why LLMs are unreliable at math, and when they need to be paired with other tools. Glenn says you should never rely on an LLM by itself to build something like an amortization table. AI performs calculations correctly more often when writing and running code, like Python.
  
  
• How to create auditable trails, including logging prompts, outputs, and assumptions. Everything AI produces must be repeatable and defensible, which means storing responses and being able to show exactly how a number was generated.

“You can’t tell an auditor, ‘the number came from the black box’,” Glenn says.

The goal isn’t to turn finance professionals into data scientists. It’s to make sure that when AI is used in forecasting, variance analysis, or reporting, leaders understand what it’s doing well, what it’s bad at, and how to defend the output.

3. Invest in process documentation

For years, SOPs have been treated like internal hygiene. In 2026, they’re training material.

“Those SOPs could turn into what we’re using to train AI agents, not just humans,” Glenn says.

This means:

• Reviewing how work actually gets done, not how it’s described in policy docs. The work people actually do – including workarounds and informal steps – matters more than how the process is supposed to work on paper, because AI can only be trained on what’s really happening.
  
  
• Documenting decision paths, approvals, and handoffs. You’ll still need human oversight when agents take on routine work, so make sure you understand where these touchpoints live.
  
  
• Identifying where judgment versus rules apply. Finance processes often mix structured rules with human judgment. Be clear about what can be standardized and what requires discretionary decisions.

AI can only automate what’s explicit. Process clarity is what allows finance teams to safely delegate work to AI while retaining control and accountability.

4. Invest in internal AI champions with domain knowledge

Rather than hiring outside AI experts, some finance organizations have promoted their champions into AI roles and backfilled their original positions – with great success.

“These people already know the systems, the data, and the business context,” Glenn says.

Their value isn’t necessarily technical depth, it's translation. They can more quickly apply company standards, adapt and redesign familiar workflows, and verify acceptable outputs.

And they’re already known and trusted in the organization – this makes them a crucial lever in your change management strategy. Their guidance is more likely to be trusted because it comes from someone who understands the day-to-day realities of the job, and they won’t be perceived as theoretical experts parachuting in but rather a peer who found a better way to work.

Why the fundamentals should be your 2026 strategy

As a former CFO, Glenn understands the instinct to look for transformational ROI. But without investing in these big fundamentals – that admittedly don’t have a shiny payoff – you’re ultimately paying for the price of your shortcuts.

Data is the area finance leaders are loath to fund, but it’s essential to your 2026 AI strategy.

“I was talking to a CFO the other day and he said, ‘I’m not spending a nickel on data’. And I thought, ‘I don’t know why we’re having this conversation then’,” Glenn says.

He sees this with a lot of the investments that bring soft ROI, but they’re the necessary foundation for harder ROI. Without process clarity, unified data, and safe AI skills, finance teams won’t unlock AI’s potential.

“This is the clarion call,” Glenn says. “If your data and processes aren’t in order, you will get left behind.”