The Data Moat Is Getting an AI Upgrade

In payments, data has long been treated as a strategic asset, a moat to defend and deepen over time.

“Every company has a data moat,” Ahsan Shah, SVP of analytics and AI at Billtrust, told PYMNTS during a discussion for the April edition of the “What’s Next in Payments” series, “The Data Game.”

But as artificial intelligence reshapes how organizations extract value, that moat is no longer defined by sheer volume. Instead, a new layer is emerging above it, one that’s built on context, metadata and domain intelligence.

“Data is the oil. And we’ve heard this for decades now. But now with AI, I think the game is changing,” Shah said. “What’s the point of having terabytes of data when you can’t actually tap into it?”

What differentiates leading companies today is not just access to data, but the ability to interpret and operationalize it. The shift is a structural one, and it’s already forcing companies to rethink everything from infrastructure to customer experience.

“Accumulating data is great,” Shah said. “But it’s not going to make the big difference. The untapped value is: Do you have a context layer? Do you have metadata? Do you have domain layers that differentiate your business?”

Rise of the Context Layer

Data alone is inert, but context makes it actionable. The “context layer” represents a structured overlay of metadata, governance, and domain-specific intelligence that can allow AI systems to understand and act on enterprise information.

“Can AI understand your data? What is the context? How do you govern it? How do you put guardrails? That’s the new ball game,” Shah said.

The implications are already visible in how companies approach analytics. The static dashboards and one-off insights of traditional reporting are giving way to dynamic, AI-driven reasoning systems.

“We’re allowing AI to say not just what happened, but why did it occur,” Shah said, noting the result is not just efficiency, but velocity, with entire feedback loops from customer input to product iteration becoming AI-native.

“Something that would’ve taken three months to build is taking three days to build,” he said, pointing to AI-assisted coding, agent-based workflows and automated product development cycles.

Nowhere is this transformation more evident than in customer experience. Payments, historically governed by rigid rules and static processes, are becoming fluid and adaptive. Instead of applying broad policies, companies can tailor decisions in real time, such as offering early payment discounts to specific buyer segments, adjusting risk thresholds dynamically, or automating collections strategies based on behavioral patterns.

“We are not at the point where it is a dark factory, with just robots,” Shah said. “You still have human in the loop.”

But the balance is shifting. AI is increasingly handling analysis and recommendation, while humans focus on judgment and oversight.

Expansion of the Data Universe

Underlying this shift is a dramatic expansion in what counts as usable data. Structured transaction records are now only part of the picture. Unstructured inputs like emails, call transcripts, invoices, CRM notes and more are being integrated into a unified intelligence layer.

“We used to say, ‘You’ve got to conform to these six things.’ Now, we’re getting into: Give us the ocean,” Shah said. “The diversity of data alongside this AI layer of context is the most powerful asset for any company.”

With AI capable of parsing and synthesizing vast, heterogeneous datasets, companies are no longer forced into rigid schemas. Instead, they can embrace complexity and extract value from it.

“Data management strategies for enterprises are the foundation of everything,” Shah said. “If you have bad data, you won’t get very far.”

Looking ahead, Shah sees the next frontier in “agent harnesses,” or systems that orchestrate multiple AI agents to execute complex workflows autonomously. These systems go beyond assistance. They represent a shift toward delegation, where AI not only supports tasks but owns entire processes, from design to execution.

“I’m going to give you everything you need, and I’m going to leave,” he said of the upcoming technological paradigm. “By the time I come back, I want you to be done.”

It’s an “amazing unlock,” he said, enabling teams to build multiple features simultaneously and dramatically increase output. But it also introduces new challenges around governance, cost, and reliability. “What won’t work is to say, ‘We don’t need human beings at all — just hit a button.’”

“You have to have guardrails. You have to have enterprise security,” Shah said. “You have to think about storage, token costs, ROI.”

But for enterprises navigating this transition, the message is clear: AI is no longer optional.

“In the last six to 12 months, there’s no debate now. AI is about survival, it’s existential,” Shah said.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.