top of page

The Three AI Questions Every AP Leader Should Be Asking

3 min

By: Daniel Shore

Screenshot 2026-08-18 at 1.05.55 PM.png

Every week you'll hear someone describe an AP platform as "AI-powered." It could be during a a webinar, a conference presentation, an analyst briefing, or even an internal discussion about your AI roadmap.

The problem is that "AI-powered" has become more of a bumper sticker than proof of capability.

That's not to say AI isn't changing Accounts Payable. It absolutely is. But "AI-powered" now describes everything from OCR, intelligent workflow routing to transformative AI native architecture.

The future of AP isn't faster exception handling. It's fewer exceptions to begin with.

Those are very different approaches with very different outcomes. Before you invest another hour investigating AI or AP technology, ask these three questions. They'll tell you far more than the vendor ever will.

1. Is AI helping us process exceptions—or prevent them?
 

Most AP automation projects have already solved the easy invoices. The remaining work—the exceptions—is where AP teams spend most of their time. Price discrepancies. Quantity mismatches. Missing receipts. Supplier questions. Most AI today helps classify, route, or prioritize those exceptions once they've already entered the process. That's useful.
 

But a more important question is: Can the platform learn why these exceptions keep happening and help prevent them before they reach AP?
 

The future of AP isn't faster exception handling. It's fewer exceptions to begin with.

 

2. Does AI understand the entire process—or just one task?
 

Many platforms have added AI to individual functions:

  • OCR

  • Coding suggestions

  • Approval routing

  • Invoice classification
     

Each improves a specific step. But AP doesn't operate one step at a time. Every invoice is connected to purchasing, receiving, supplier communications, contracts, payments, and finance. The best AI doesn't simply automate individual tasks. It understands the relationships between them.
 

Ask your vendor: Where does AI make decisions across the process—not just inside one workflow?
Architecture matters because context matters.

 

3. Does AI include suppliers—or only AP?
 

This may be the most overlooked question of all. Most invoice exceptions don't begin inside Accounts Payable. They begin before the invoice is submitted—or because suppliers lack the information they need to resolve issues themselves.
 

During a recent supplier interview, Abhishek Bhardwaj, Sr. Administrator at Kraft Heinz, described the experience before using an AI enabled Supplier Portal:
 

"Before checking the portal, the alternative was reaching out to Gordon Food Service's AP team directly for every invoice status question. It's a great experience…it has all the options I need on a daily basis."
 

The lesson is simple: when suppliers have visibility into invoice status, PO depletion, schedule-to-pay dates, and dispute information, they solve many routine issues themselves instead of creating additional work for AP.
 

Ask: How does AI help suppliers submit cleaner invoices—not just help AP process broken ones faster?

That's where many organizations still have their greatest opportunity.
 

 

A Better Conversation About AI
 

The question isn't whether a platform uses AI.Eventually, they all will. The better question is:

Where does AI create better decisions? Does it reduce recurring exceptions? Does it improve supplier collaboration? Does it learn from prior outcomes? Does it improve the quality of the process over time?
 

Those are the conversations worth having. Because the goal isn't simply to make AP more efficient. It's to build a process that gets smarter every month.

 

Final Thought
 

The next time a vendor tells you their platform is "AI-powered," don't ask what AI features they have. Ask what problems their AI solves—and whether it's designed to eliminate tomorrow's exceptions instead of simply processing today's. That's the distinction that matters.

 

Continue the Conversation
 

How is your organization evaluating AI in AP? Which of these three questions is receiving the least attention today?
 

I'd like to hear your perspective as we continue exploring what AI-native finance operations should look like. Reach out. 

Daniel Shore
sweater.png
decemberdesk.png
bottom of page