AI in Finance: What Leaders Are Actually Doing to Build a Smarter AP Function

Accounts payable has always carried operational risk. It’s the function that moves the money, sitting at the intersection of fraud exposure, compliance pressure, and rising business pressure from CFOs and boards. What’s changed is the scale of that risk and the expectations around how it gets managed. The old playbook of more manual reviews, more approval layers, and more headcount doesn’t hold up when transaction volumes are growing faster than teams can scale. That’s why AI-enabled accounts payable automation has moved from a nice-to-have to a strategic priority.

That tension was the starting point for a recent webinar I joined with Mark Brousseau, President of Brousseau & Associates, and Lisa Ricardo, VP of Enterprise Performance and Technology at CBIZ. We covered where AI in finance is delivering real results in AP today, what governance needs to look like before you deploy it, and what it means in practice to keep humans in the loop. Below are my key takeaways.

The pressure on AP teams is real, and it’s compounding

When we asked attendees what their biggest challenge was, 44% identified implementing new technologies effectively as their top concern, with talent shortages and bandwidth constraints coming in second at 17%. Those two things are connected. Teams are being asked to handle growing transaction volumes with fewer people and are looking to technology to close the gap. The risk is doing that without inadvertently creating new problems in the process.

Mark framed it well: AP today sits at the intersection of increasingly sophisticated fraud, more complex transaction flows, and rising expectations from leadership. Adding more manual checkpoints worked when volumes were manageable, but the only sustainable path forward now is embracing finance automation and changing the underlying model rather than layering more processes on top of one that wasn’t built for this scale.

What AI is actually doing in AP, and where it matters most

Software facilitates a workflow. AI-powered accounts payable automation performs the work inside it. In AP, that difference shows up in invoice capture and coding, PO matching, duplicate bill detection, approval routing, and audit trail maintenance—tasks that used to require someone to touch every transaction. AI handles them at volume, flags exceptions for human review, and produces the documentation that controllers and CFOs need when questions arise.

What AI changes is how that risk is identified. It’s monitored, and it’s controlled. Perhaps most excitedly, it does it in a continuous fashion.

Mark Brousseau, President, Brousseau & Associates

Lisa highlighted where the risk shift is most meaningful: anomaly detection across full transaction volumes rather than samples. In a high-volume AP environment, catching something unusual used to mean reviewing a slice and hoping any problems fell within it. AI now analyzes transactions, so unusual activity is far less likely to go undetected simply because no one happened to look at the right row on the right day.

Governance has to come before deployment, not after

Lisa made the case that deploying AI in accounting and finance without a governance framework is one of the biggest risks finance teams are taking right now. The barrier to entry for building something that touches your trial balance has dropped: a team member can connect an AI tool to your ERP with natural language prompts and create something that generates journal entries or changes how financial data gets interpreted. Without a human-in-the-loop process for verifying that output is accurate and auditable, that’s a significant exposure.

Giving everybody an AI platform and just letting them go wild is like giving a toddler a flamethrower.

 Lisa Ricardo, VP Enterprise Performance & Technology, CBIZ

The point isn’t to slow adoption down. It’s to treat it like any other technology rollout: understand your current landscape, define a governance policy, and start with something contained. Prove it works, verify it’s auditable, then expand.

Start with the problem. The technology follows.

When Mark asked for the single most important piece of advice for teams struggling with implementation, the answer was consistent: get specific about what you’re trying to solve before evaluating any technology. Are you trying to reduce fraud exposure? Close the books faster? Stop adding headcount every time invoice volume grows? Each one points to a different starting place and a different measure of success. Tools that don’t deliver ROI usually weren’t pointed at the right problem to begin with.

• Define what you’re trying to fix before talking to any vendor, and let those criteria drive the evaluation rather than the demo.
• Build your governance policy before anything goes into production, not after you’ve discovered a gap.
• Start small, prove it works, and expand from there.

Watch the full session

The full webinar goes deeper on how AI in finance is reshaping AP operations, including a live Q&A with real questions from finance leaders working through AI adoption right now. If scaling your AP function with AI is on your agenda this year, this session is worth a watch.