The finance technology playbook was once relatively straightforward. CFOs built out their department tech stacks by putting one piece of software after another between an employee and a cumbersome business process.
For many teams today, that’s still the playbook. Accounts receivable teams have embraced dashboards, collections teams use workflow tools, credit teams rely on decision engines, and payments teams are still logging into portals every day.
But Dave Ruda, vice president, product, at Billtrust, told PYMNTS for the September edition of the “What’s Next in Payments Series: The Fall Draft” that agentic artificial intelligence isn’t just challenging that legacy architecture, it’s already breaking it apart.
“Think about all the tasks you do now being spun up by agents and being automatically executed,” Ruda said. “That’s pretty powerful and very real.”
The next big interface for enterprise payments, it turns out may be no interface at all. And that is a bigger shift than automating one more AR workflow. It potentially changes what enterprise financial software is for, and what the people using it do.
Agentic Payments Mean the Front End Starts to Matter Less
The payment landscape is starting down a path toward financial systems that employees themselves are no longer required to operate directly. AI agents are capable of completing tasks such as being asked to retrieve data, interact with payment infrastructure and eventually execute routine tasks across enterprise systems. This positioned humans as strategic oversight managers responsible for supervising exceptions rather than pushing every transaction through the machinery themselves.
“There is a world where there’s less human in the loop, where you’re only reviewing exceptions, and the norm is handled by the robotics because it can and it can do it just as well,” Ruda said.
One enabling technology is Model Context Protocol, or MCP, which provides a standardized way for AI applications to connect with external tools and data. The significance for payments is not merely better integration. It is that agents can potentially interact directly with enterprise systems rather than requiring employees to navigate their interfaces. An employee might ask an AI system for information or give it an instruction, while the agent handles the interaction with the underlying financial application.
“You don’t need front end to do anything. You just need to be able to talk to the agent,” Ruda said, describing the emerging model as “headless” financial software.
For CFOs, that changes the AI question. The strategic issue is no longer simply how much generative AI can improve employee productivity. It is which financial workflows still require an employee to operate software at all. The employee does not disappear. The job moves up a level: defining rules, monitoring systems, handling anomalies and intervening when machines encounter something outside the norm.
Getting the Plumbing in Place Needs to Come Before Autonomy
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
There is a catch. The more autonomous finance becomes, the more important its decidedly unglamorous infrastructure becomes.
Asked where businesses should be placing their bets, Ruda started not with AI but with global invoicing compliance. Companies expanding across borders face an complicated collection of electronic invoicing and tax requirements. Those rules determine how invoices are created, transmitted and registered with government authorities.
“My first pick though is the global compliance piece,” Ruda said, explaining that AI may end up increasing the value of the plumbing it is supposedly leapfrogging. “If you can’t get that done, then you’re not going to make it to the second round.”
After all, autonomous financial systems cannot reliably execute transactions if the transactions themselves do not comply with the rules governing them. The further companies remove humans from routine execution, the more important those machine-readable rules and controls become.
That’s also why Ruda argued that finance needs to move beyond the familiar concept of “black box” AI toward what he called “glass box AI.” That distinction becomes more important as AI progresses from recommending actions to executing them.
“You can see everything that’s happening in the box,” he said. “It’s all safe in there. It’s protected and it’s manipulable.”
An opaque model generating marketing copy creates one category of risk. An opaque model making credit, collections or payment decisions creates another.
Ruda’s test is considerably simpler than most enterprise AI governance frameworks: “Would you bet your job on that?”
If not, the system probably has not earned the authority being handed to it.
Watch the full PYMNTS TV interview with Billtrust’s Dave Ruda to hear more about:
- Why AI could make the financial software interface disappear. Ruda says MCP-style connectivity could let agents interact directly with payments systems, turning today’s dashboards and portals into “headless” infrastructure where AI retrieves data and eventually executes routine financial tasks.
- Why the AI vendors with the best demos may not survive finance’s trust test. Ruda warns CFOs not to get caught up in the “jazz hands,” arguing that payments needs “glass box AI” where users can understand and control what algorithms are doing before trusting them with credit, collections or payment decisions.
- Why human-in-the-loop AI may ultimately be training humans out of the loop. Ruda says today’s approvals, corrections and oversight are essential to training more autonomous systems, potentially creating a future where “you’re only reviewing exceptions” while routine financial work is handled automatically.
Leave a Reply