And to answer that question to the best of their abilities, lenders poured investment into underwriting models, alternative data and faster origination systems.

But that era is no longer where the most important competition is happening.

“The frontier has moved,” Ryan Dew, chief product officer at Thredd, told PYMNTS. “The innovation is happening at the point of transaction.”

This shift is being driven by the explosive growth in contextual data embedded in every payment. Transactions today are no longer just amounts and merchant codes; they carry behavioral signals, device data, and, increasingly, contextual inputs from emerging forms of agentic commerce.

“The data that is now embedded in the transaction itself is much richer than it ever has been,” Dew said.

Static rules and pre-set underwriting decisions can struggle to keep pace in this environment.

See also: The ABCs of AI Credit: A Playbook for Issuers

Credit Moves From Static Decisions to Streaming Intelligence

The result is a new operating model for issuers. Credit risk is no longer managed only at origination; it is increasingly interpreted continuously, as customer behavior changes.

“When you start talking credit, it’s all about how do issuers manage risk, right? And that’s always been the name of the game,” Dew said, pointing to a new, and increasingly important metric, “time to signal,” or “how long is it taking to act on a change in customer behavior.”

In a market where speed is becoming a key determinant of performance, the firms that can detect, process and act on signals fastest may gain a competitive edge. Artificial intelligence (AI) is becoming central to this acceleration and reshaping the role of human decision-makers.

“The humans are not necessarily the ones that are monitoring the transactions,” Dew said. “They’re the ones that are designing the system that write the rules.”

A fraud rule that might once have been manually coded years ago can now be generated, updated and refined by machine learning systems trained and overseen by humans.

“There’s nothing that’s off limits to try to automate,” Dew said, noting ongoing pressures for firms to reduce costs and improve efficiency.

“But the one thing technology can’t automate is the legal architecture,” he added, referring to compliance obligations and consumer protections that anchor financial decision-making.

This has created a bifurcated model of automation where on one side are “machine-owned” processes made up of high-volume, low-risk and reversible decisions that lend themselves to automation. On the other are “human-owned” processes tied to regulatory obligations and consumer rights, where full automation remains elusive.

Where Competitive Advantage Shifts Next

The same architectural problem appears in the industry’s long pursuit of a single view of customer risk. Financial institutions have historically relied on a patchwork of point solutions, each optimized for a specific function or payment rail. The result is a fragmented data landscape that resists consolidation.

Dew argued this is “not a technology problem” so much as “an architectural problem.” Point solutions may understand fragments of a customer’s behavior, but issuers increasingly need consolidated platforms that can see across card rails, account-to-account payments, cross-border transfers and other flows.

What is changing now is the pressure to unify these systems, driven by the need for real-time decisioning. As transaction data becomes the primary source of insight, firms are increasingly focused on consolidating their service stacks.

“You’re starting to see providers try to consolidate those so that they can get that 360 view of the customer,” Dew said.

After all, consumers themselves are no longer satisfied with static financial products. They want dynamic, personalized experiences that adapt to their behavior in real time. As a result, the transaction itself is where consolidation becomes most valuable.

“The most important part in the customer journey always is at the point of transaction,” Dew said, because that is where issuers can see how customers are actually behaving, spending and interacting with financial products.

That has implications for the future of credit products themselves. Customers may not merely want access to multiple accounts or repayment options; they may expect software to make contextual choices for them in real time.

“Anyone who can automate that in a way that is very tailored to the specific customer and their use case needs is ultimately who wins,” Dew said, flagging the emerging opportunity in systems that can learn how a customer wants to use financial products and automate those preferences inside the transaction flow.

“It all comes together in the transaction,” he added. “Making it programmable is where the innovation is.”