In a time when buzzwords like “agentic AI” and “autonomous systems” are being tossed around with Silicon Valley bravado, the B2B payments space is looking for results. Not just hype.
At the same time, as digital payments accelerate by the millisecond, artificial intelligence (AI) is no longer a futuristic concept for B2B firms but has fast become an operating necessity. Its latest flavor? Agentic AI: autonomous software agents that don’t just assist, but act.
“I think we’re standing at the edge of a major transformation,” Coupa Product Strategy and Management SVP Rajiv Ramachandran told PYMNTS during a conversation for the June “What’s Next in Payments: Secret Agents” series.
“Agentic AI is not just a technological trend — it’s a rethinking of how decision-making, risk and value creation happen inside financial workflows.”
While the term “agentic AI” may sound futuristic or like something out of a Cold War spy thriller, Ramachandran said he sees it as a pragmatic solution to an old business imperative, that of doing more, faster and more securely, with less human overhead.
“It’s not just about automation; it’s about delegation,” he said.
In payments, this means software that not only processes invoices or flags fraud but systems that can actually decide whether to move money, when to move it, how much and to whom. All while checking for compliance, validating identities and maintaining audit trails.
It’s the kind of operational intelligence that’s been sorely missing in the fragmented, friction-heavy world of B2B commerce.
Payments Security Is the Front Line
Agentic AI, on a purely technical basis, represents a profound leap from traditional automation. These systems aren’t just rules-based assistants. They observe, learn, predict and act.
The promise of agentic AI lies in its ability to collapse complex, manual workflows into intelligent, streamlined processes, and Ramachandran noted that the impact of agentic AI could be as significant as the shift from paper to digital was a generation ago.
Still, to enable agentic AI, systems need access to clean, structured data and real-time protocols. For example, being able to validate a bank account instantly, check it against fraud patterns, and execute payments without manual formatting errors.
“You need a seamless data model,” Ramachandran said. “The agents can’t act unless they’re acting on trustworthy, synchronized data. That’s job number one.”
When asked where agentic AI might first gain traction in the payments ecosystem, Ramachandran didn’t hesitate. “If I were to summarize,” he said, “security, autonomous payments, and analytics and insights. Those would be my top three.”
Ramachandran pointed to the plague of business email compromise (BEC) attacks, where hackers impersonate suppliers and redirect funds, as an area where agentic AI can help stop fraud before it starts.
“You’d be surprised how often people fall for these things. It’s a simple email, and suddenly the payment’s gone,” he said.
But an AI agent, operating in real time, could verify banking details, assess account legitimacy and flag inconsistencies before a human ever clicks “send.” After all, in many cases, fraud often starts long before the actual payment.
“People think about security as payment security only. But fraud can and does happen way upstream in the process, at the purchase order or invoice level. If you don’t check those documents for integrity, you’re just trusting bad data further down the chain,” Ramachandran said.
“We understand who the risky suppliers are, and we understand the risky transactions. That’s been the foundation of a lot of our own capabilities,” he added, noting that Coupa’s holistic approach has yielded products like Spend Guard and Risk Aware, built atop a proprietary dataset of trillions of dollars in B2B transactions.
Autonomy and Intelligence
If security is about mitigating risk, automation in B2B through agentic AI is about unleashing potential.
“Today, payments are siloed, full of paper, handoffs, manual approvals,” Ramachandran said. “But if you truly have digital documents all the way from purchase to payment, there’s no reason those can’t be connected by an autonomous, intelligent process.”
This isn’t just about efficiency. Autonomous payment execution has strategic implications for liquidity management, supplier relationships and cross-border trade. Ramachandran believes that with the right infrastructure, including secure data models and real-time APIs, AI agents can manage billions in transactions without manual oversight.
“It’s our belief that in our network, in the future, we are going to see more transactions autonomously done rather than manually done,” he predicted.
Still, the most recurring theme in Ramachandran’s vision for agentic AI isn’t intelligence or efficiency. It’s trust. Without it, he insisted, autonomy is just a theory.
“It’s extremely important for us to earn the trust of that customer and that end user so they can actually execute these use cases in a streamlined and autonomous manner,” Ramachandran said.
