The modernization of banks’ technology stacks, to this point, have only been the initial steps in a larger process. The adoption of real-time payments has shown how risk, compliance, and customer experience need to catch up to the instant payment environment. Even less visible functions like treasury services are being touched by modernization. These changes are blurring the lines between platforms and processes as accountability continues to shift among banks, tech vendors, fintech partners, and third-party providers.
The 2026 Tech and Infrastructure Trends report from Javelin Strategy & Research looks at how these trends will play out over the coming year and beyond. Complicating matters is the arrival of artificial intelligence, which could play a key role in how payment technologies evolve.
Moving Past Legacy Technology
Bank modernization exists along a spectrum. Some have already added capabilities for products like real-time payments and instant fraud detection. At the other end, many banks still using their legacy cores have cobbled together solutions through middleware that allows them to offer these capabilities, but not in the most efficient or best way.
The biggest bottleneck for most banks is legacy technology. If they have not yet made progress in modernizing their systems, they will have issues.
Legacy cores still work in many ways because many payments, in the macro sense, still do not need to be processed instantly. It doesn’t necessarily matter to a typical retail customer if the bank offers instant settlement or instant authentication; they just care that their bill is being paid.
Batch processing allowed banks to methodically look through transactions for fraud and other risks, which made mitigating and remedying such issues much easier. But the era of instant payments created a disconnect. A small business prefers to pay its suppliers as late as possible, and to do that the business will want to do it instantly. If the bank’s core can’t handle that, it is likely to make the solution the institution does put together a little more clunky or more difficult to handle.
Different Risks for Different Banks
This scenario presents different risks for different types of banks. Bigger banks have the financial resources and highly skilled personnel capable of adding these capabilities and staying on the leading edge of next-generation payment technology.
“Look at a bank like JPMorgan, which has always stayed one step ahead,” said Matthew Gaughan, Payments Analyst at Javelin and a co-author of the report. “Now they’re trying to emulate the fintechs through having their own developer portal, working with blockchain technology, trying to stay one step ahead and build a foundation for future payment technology. The bigger banks have an easier time being proactive and figuring out what makes sense for them, which technologies they want to implement, and also are able to hire the talent to do so.”
On the flip side, smaller banks don’t have as much money and may not have the ability to pull in the necessary talent. There is also a big group in the middle made up of banks that are able to utilize core banking providers and process payments as well. Fintechs like Fiserv and Jack Henry are rolling out more modern payment platforms that have real-time payments and an infrastructure more capable of orchestrating these payments, so these small and mid-sized banks do have access to these services.
Taking Responsibility for Fraud
Real-time payments also mean real-time fraud, and banks will need to be able to manage both. Agentic commerce complicates matters by opening up new vectors for bad actors to use the same technology to their advantage. It is still unclear how that risk will be assigned.
“If you’re still relying on batch processing, you’re obviously not going to be able to offer real-time payments,” Gaughan said. “But your fraud tools and capabilities are probably also lacking and would not able to properly monitor and mitigate fraudulent activity happening in real time. That makes it harder to remedy these situations.”
Who takes responsibility in an increasingly fragmented payment environment? Companies like OpenAI, Mastercard, and Visa have established agentic payment protocols that do the legwork for the banks, but does that make them ultimately responsible for the success of the payment?
OpenAI’s agentic commerce protocol shows how the new landscape might work. Co-developed with Stripe, it essentially acts as a shared language among the AI agent, the merchant, and the bank. All that transaction information is used to create a payment token. The merchant processes the payment in the typical manner, using its payment service provider.
In the end, OpenAI is providing the plumbing in that shared language and not necessarily doing anything directly with the payment. That has created a gray area around who’s responsible for fraud or errors.
“Banks have their own APIs that they’re opening up to third-party developers,” Gaughan said. “The bank is processing the payment, but it’s happening on this fintech’s platform or this technology company’s platform. Who’s responsible for making the customer whole? All these things are opening up these new questions, and we’re still in the early stages of figuring it out.”
Opportunities for Treasury Services
For banks seeking to modernize their treasury services, the challenge is that much of the data and processes for such services remain siloed. There may be a possibility for AI to be used in this process, but centralizing the data in those processes is the foremost concern. So much of the more modern technology relies on such data, which means it needs to be accessible, readable, and digestible. That will allow banks to automate different treasury processes and solutions.
“Treasury services solutions are almost a perfect fit for things like large language models, because they provide very clear processes,” Gaughan said. “It’s not as overly complicated as some other parts of the payment landscape. They are data-centric, rules-based, and fairly consistent, and that’s just the type of information that a third party or proprietary LLM would thrive on. A business owner who wants historical data on accounts receivable days could quickly visualize it as a chart, for example. That’s crucial information for merchant clients who are looking to see what where their liquidity is at or maybe better optimize their cash conversion cycle and keep more of that money for themselves for longer.”
Certainly, artificial intelligence will have an impact on all parts of the payment process. One aspect that still lies in the future is the potential for bad actors to game the system, perhaps by creating fraudulent bots that can complete transactions.
As Gaughan said, “It’s an interesting time.”
