The age of agentic AI is here, and it’s rewriting the rules of financial services.
Artificial intelligence (AI) is experiencing a fundamental shift. Once confined to the realms of content generation and isolated automation tasks, AI’s capabilities have advanced beyond its game generative capabilities toward what many industry leaders call “agentic AI.”
Characterized by the ability to reason, learn, make decisions and act autonomously, agentic AI software solutions are being rapidly embraced by payments and financial services companies looking to transform their operations. However, while the technology’s promise is enormous, success may depend on how effectively these companies integrate AI into their existing infrastructures and workflows.
A deep dive into recent conversations with payments industry leaders for “What’s Next In Payments Series: The Rise of Digital Labor: Exploring Agentic AI in Banking and FinTechs” reveals four defining themes for how the marketplace is thinking about the impact — and the integration — of agentic AI across payments:
Agentic AI holds transformative potential for the payments and financial services industries. However, as these executives illustrate, the companies that invest in building the right foundation today will likely be the leaders of tomorrow.
Agentic AI Is Already Having an Impact in Payments and Financial Services
Agentic AI is already proving its worth across the industry. Companies are deploying these systems to optimize processes such as fraud detection, compliance monitoring and cross-border payments. Instead of merely flagging anomalies, agentic AI can autonomously initiate investigations, gather relevant data and even propose solutions.
It’s a shift that promises to reduce operational burdens and enhance accuracy, saving both time and money.
For Mark Sundt, chief technology officer at Stax Payments, previous AI models were “a mile wide and an inch deep.”
He said that the introduction of the Model Context Protocol (MCP) is nothing short of revolutionary. Sundt described MCP as “the lingua franca for these models to communicate back and forth — and to ‘do’ discover and expand their capabilities on the fly.”
He added: “It’s absolutely game changing … and produces better outcomes than previously manual efforts.”
Back-Office Infrastructure and Data Readiness Are Critical Enablers
Despite the tremendous potential of agentic AI, executives stress that its effectiveness hinges on the quality of the underlying infrastructure. Robust, integrated, high-quality data ecosystems are essential. Without them, even the most advanced AI models will struggle to achieve meaningful results.
i2c CEO Amir Wain believes that many financial institutions are still only scratching the surface of what agentic AI can achieve. “The tools themselves are ready, but the underlying infrastructure within financial institutions is often ill-equipped to serve up the necessary data fluidly and consistently,” PYMNTS wrote.
Whether tackling complex compliance requirements or enhancing transaction security, data quality directly impacts the performance and reliability of AI-driven solutions. Only with good data can AI systems extract meaningful insights and act autonomously with confidence.
“The models are only as good as the data being fed to them,” Rinku Sharma, chief technology officer at Boost Payment Solutions, told PYMNTS. “Garbage in, garbage out holds even with agentic AI.”
Realizing Agentic AI’s Capacity for Scaling and Automating Processes
Agentic AI systems often employ technologies such as reinforcement learning, natural language processing (NLP) and deep learning to achieve a high degree of autonomy. By leveraging these tools, agentic AI can process vast amounts of data, identify patterns and make decisions that would be impractical or impossible for human operators to achieve in real time.
This can make them invaluable across enterprise back offices.
Unlike traditional automation, which relies on rigid, predefined workflows, agentic AI can adapt its approach based on situational context. Instead of merely flagging anomalies, agentic AI can autonomously initiate investigations, gather relevant data and even propose solutions. It’s a shift that promises to reduce operational burdens and enhance accuracy, saving both time and money.
At TerraPay, Co-Founder and COO Ram Sundaram is working to use AI models to streamline cross-border payments. The company’s infrastructure connects 3.7 billion mobile wallets across 200 sending and 144 receiving countries. AI models and agentic AI “are extensions of what we’ve always valued at TerraPay,” he told PYMNTS, emphasizing that efficiency, security and affordability remain paramount.
Sharma told PYMNTS that Boost is making “giant strides” in embedding AI into day-to-day operational procedures, particularly across areas like merchant onboarding.
“It’s about intelligently acting on the data, taking decisions and creating those automated workflows that create the scale that is needed,” he said.
Embracing Experimentation’s Role in Driving Customer-Centric Innovation
Innovation is seldom a straight line, and executives are approaching the deployment of agentic AI with careful strategy and playful experimentation.
For WEX Chief Digital Officer Karen Stroup, the best approach to deploying agentic AI involves a disciplined strategy of experimentation. “If you’re going to experiment with agentic AI or any type of AI solutions, you want to focus on two things. One is the area where you’re most likely to have success. And two, is there going to be a good return on that investment?”
And in a competitive marketplace, businesses are experimenting with agentic AI to enhance the customer experience. From identifying customer needs to offering tailored solutions, agentic AI is helping companies anticipate preferences and deliver frictionless interactions.
TerraPay is using AI models with machine learning to bolster customer support and automate tasks as financial institutions (TerraPay’s client base) send payments in real time, and those payments are processed into local markets’ beneficiary banks.
i2c’s Wain highlighted the importance of a “one unified customer” model.
“Even the best AI models perform suboptimally when data is segmented by product or siloed in disparate systems,” Wain said.
According to him, the aim should be to create a unified view that spans all relevant services, such as checking accounts, credit cards and loans, so that the AI can draw upon a complete picture of customer needs.
Ultimately, realizing the potential of agentic AI requires a holistic approach that balances technological innovation with ethical considerations. Successful adoption depends on robust infrastructure, high-quality data, iterative experimentation and a willingness to push beyond conventional workflows. Only then can agentic AI revolutionize how processes are scaled.
