During an episode of The Treasury Whisperer’s Talkcast Podcast, Better Together: When Banks and Fintechs Collaborate, Huntington National Bank’s Deepak Kapoor joined Alacriti’s Stuart Bain to discuss how banks are helping commercial customers modernize payments. The conversation explored everything from embedded banking and receivables modernization to AI, instant payments, and the future of agentic commerce. Here are some of the key takeaways from that discussion.
Modernizing Payments Means Modernizing Workflows
One of the strongest themes throughout the discussion was that modernization begins well before a payment is processed. “It’s for various reasons in the U.S., people just love checks,” Kapoor said. Rather than simply encouraging customers to adopt new payment rails, Huntington focuses on becoming part of the customer’s workflow.
“A lot of our effort is to go up front in the workflow and do more on that side. This is where, for example, an invoice is presented when the EBPP platform comes in saying, ‘Hey, now I’m in your workflow. My systems are part of your workflow, and now it’s easier for you to adopt the payment methods that we’re suggesting.’ I think that’s one that’s one big strategy, outside of just educating customers around the cost of fraud and delays, and obviously collecting faster,” Kapoor said.
Bain explained that this philosophy is reflected in Alacriti’s invoice presentment capabilities, which have evolved to support commercial invoice payments as organizations look for better ways to digitize receivables.
Integration Is Essential
Kapoor noted that commercial customers increasingly expect their banking services to connect directly with their existing systems. “I think we are in what we call an integrated economy. People call it embedded payments, embedded banking. You can’t be in a customer’s workflow if you’re not connected to their systems,” Kapoor said. He explained that Huntington has invested heavily in connecting with ERP platforms, treasury management systems, and customer workflows while also partnering with providers like Alacriti to build custom integrations for commercial clients.
Better Data Creates Better Experiences
Moving money is only one part of the payment process. “If we don’t provide the reconciliation data, then those payments end up in a suspense account from a GL perspective. Our customers have teams whose only job is to do recon. ‘Where is my cash balance? Where’s my payment status? Did this invoice get paid or not?’ Getting rich payment data back is very high priority,” Kapoor said.
Bain explained that Alacriti separates payment data from settlement so organizations receive richer information alongside each payment. “We’ve basically taken that approach that the data is one thing and the dollars are another,” Bain said.
AI Is About Making Operations Simpler
Both speakers agreed that AI’s greatest near-term opportunity is improving operational efficiency. “If your job is recon, you come in the morning and just say, ‘Hey, help me reconcile,’ and it just reconciles. If you come in the morning and say, ‘Do my cash flow forecast or chase my cash,’ the AI is connecting with the bank and Alacriti to pull all of the data and do the job for you. In a way, AI could do for treasury functions what Claude Code does for a developer. That’s what we are seeing, and eventually that’s one of the hypotheses—that’s where it’s going to go. We’re starting to see some early signs,” Kapoor shared.
Bain described several AI initiatives Alacriti is exploring, including reporting, reconciliation assistance, document processing, and conversational payment experiences. “We’re trying to keep the scope of AI constrained,” Bain said.
Real-Time Payments Require Real-Time Operations
Because of instant payments, organizations must also rethink how they manage receivables and reconciliation. “Our approach with our clients is to build real-time integrations into the ERP systems. We actively chose to build the integrations into the cores and make them real-time. So anyway, you can do faster payments. If that’s in place, you’re all good. I can make a payment anywhere, the payment will flow in, and the system will be updated in real time,” Bain said.
Bain also noted that real-time settlement changes how organizations must think about reconciliation. “If my average payment’s $100, I’ve now got a thousand transactions settling into my DDA. Most reconciliation systems aren’t designed to scale to that. I suspect what we’re’re going to start to see is more use of things like sweep accounts and artificial end of day,” Bain said.
Kapoor added that there is no one-size-fits-all approach because organizations have different operational needs. “If you’re on the smaller side, I think the basic technology works. A lot of it is just segmenting your customers and understanding what they might need. The larger ones need a bit more tooling to really automate it. At the large end, I think their focus isn’t a liquidity deficit issue—it’s an investment strategy saying, ‘Hey, I have surplus cash, let me invest it.’ On the smaller side, it’s mostly like, ‘Hey, I need the money now.’ It’s different problems you’re trying to solve for, and hence different solutions.”
Agentic Commerce Will Require New Rules for Payments
The conversation concluded with a look at agentic commerce and what happens when AI agents begin initiating, negotiating, and completing payments on behalf of consumers and businesses. Kapoor emphasized that while the technology is advancing quickly, much of the supporting infrastructure is still evolving. “It’s early days. Lots has been done to formalize the protocols for payments, but you still have a lot of infrastructure for agent interaction being built or even being debated. The big open items are: how do you manage disputes? I don’t know if the agent had the mandate to buy something, but the agent bought something. Now you want to return it—is it a legitimate dispute or chargeback, and how do you differentiate that from fraud? There’s also the theme of how businesses are discoverable in the agentic world. Right now you have a website and a shopping cart. That doesn’t make you agentic-enabled. There’s a lot of this plumbing that has to be sorted,” Kapoor said.
Despite those challenges, Kapoor believes adoption will continue to accelerate. “I think the strong story is that consumers are trusting it. I think you’ll see a rapid explosion in the infrastructure in the next 12 to 14 months,” Kapoor said.
Bain added that banks and technology providers are already beginning to envision practical use cases while recognizing that new governance models will be needed. “We work closely with lots of digital banking partners, and we can see a real use case for agentic functions. If I receive a bill, you can pay it for me up to a certain amount. It’s like semi-automation—I don’t need to respond to a reminder and do things. But obviously there’s the liability shift. ‘I never authorized that payment, so I want my money back, but I’m keeping the goods that got delivered anyway.’ It’s going to be interesting to see how the card scheme rules, state regulations, Fed regulations, and TCH’s own rules about liability shift evolve,” Bain said.
