Kofi Ndaikate has spent his career at the intersection of emerging technology and financial regulation, navigating the complex shifts from traditional banking to the decentralized world of blockchain. As an expert in the dynamic fintech landscape, he possesses a deep understanding of how policy and infrastructure must evolve to support the next generation of digital transactions. Today, he joins Jay Thibain to discuss the emergence of agentic commerce—a transformative era where software agents take the lead in purchasing—and how established players like Visa and Thredd are building the rails to make this automated future a reality.
How can issuers integrate AI-driven payment capabilities without overhauling their entire legacy infrastructure?
The most effective way to adopt these advanced capabilities is by building on top of existing frameworks rather than attempting a total replacement of the core systems. By participating in programs like Visa’s Agentic Ready initiative, issuers can leverage Thredd’s processing platform to handle AI agent-initiated payments while keeping their underlying infrastructure intact. This approach allows firms to utilize a single API to manage debit, credit, and digital wallet functions across more than 50 countries without disrupting their current operations. It effectively creates a specialized layer where AI agents can search and recommend products, while the traditional rails continue to handle the heavy lifting of issuer approval and fraud monitoring. This ensures that the transition to agentic commerce feels like a seamless upgrade for the 100-plus fintech firms currently supported by these processing networks.
What specific security mechanisms are being developed to ensure that AI agents act strictly within the boundaries set by human users?
Security in this new era relies heavily on the concept of “agent-specific tokens” that carry very strict, tailored permissions defined by the consumer. Instead of sharing sensitive card credentials, the system uses the Visa Token Service to provide agents with payment tokens that are restricted to certain types of transactions or budget limits. To ensure that a human is still at the center of the decision-making process, biometric authentication through a Visa Payment Passkey is required to finalize the purchase. Furthermore, the infrastructure is being designed to detect “execution drift,” which is a unique behavior where an AI might start deviating from its original instructions. By monitoring for these unusual transaction patterns and using device binding, the system provides a robust safety net that protects the user even when a software agent is doing the legwork.
With established platforms like Zilch adopting these features, how do you expect the consumer relationship with credit and spending to change?
Zilch, which has grown to serve over 6 million customers since its inception in 2019, is a perfect testing ground for how agentic commerce will redefine the consumer experience. We are moving toward a model where the consumer doesn’t just use a card, but rather manages an intelligent payments platform that spans credit, installments, and earned wage access. When an AI agent is tasked with finding a product within a specific budget, it can automatically calculate which payment method—whether it’s rewards, deals, or credit building—is most beneficial for the user at that moment. This shifts the burden of financial optimization from the individual to the software, making high-level financial decisions more accessible to the average person. It is a fundamental shift in how payments work, where the focus moves from the act of paying to the quality of the purchasing decision itself.
What is the significance of using a single API and tokenization to support such a wide variety of fintech firms and digital banks?
Using a single API to provide debit, credit, and ledger capabilities is the “secret sauce” that allows for rapid, global scalability in the fintech sector. It means that whether a company is a small digital bank or a large embedded finance provider, they can all access the same high-level tools for AI-driven commerce without building proprietary bridges. This standardization is critical because it ensures that agentic commerce doesn’t become a walled garden accessible only to the largest players. By integrating device binding and passkey-based authentication into this unified API, the industry creates a consistent security standard across different markets. This allows the ecosystem to grow collectively, ensuring that as AI agents become more prevalent, the infrastructure supporting them is both universal and secure.
What is your forecast for the growth of agentic commerce over the next few years?
While we are currently in the early adoption stages, I expect the next few years to bring a massive surge in infrastructure development as processors and networks perfect the controls needed for AI transactions. We will see agentic commerce move from a niche fintech experiment to a standard feature in digital wallets, where software agents handle everything from subscription management to complex grocery shopping. As more issuers follow the lead of pioneers like Zilch, the trust gap between humans and AI agents will close, primarily because the security frameworks like tokenization and passkeys will become invisible but ubiquitous. Eventually, manual online shopping may feel as outdated as writing a physical check, as we transition to a world where our financial assistants proactively manage our economic lives with precision and speed.
