The global payments industry is quietly re-engineering itself for a future in which humans are no longer the primary actors at the checkout screen. Instead, artificial intelligence agents—software systems capable of reasoning, searching, negotiating, and executing transactions—are expected to handle everyday purchases, from booking flights to replenishing household goods. For payment giants, this shift represents a structural transformation as significant as the rise of e-commerce itself, forcing them to rethink how trust, authorization, liability, and competition function in a world where software, not people, clicks “buy.”
Executives across the sector increasingly describe this transition as inevitable. As consumers grow comfortable asking AI tools to plan trips, compare prices, and recommend products, completing the transaction within the same interface becomes the logical next step. The challenge for payment networks is ensuring that this new layer of commerce remains secure, interoperable, and economically viable.
From digital checkout to delegated decision-making
E-commerce moved payments from physical stores to screens, but it still relied on human decision-making at the point of sale. Agentic commerce goes further by delegating that decision-making itself. Instead of browsing multiple sites, a consumer can instruct an AI agent to find the best option within a defined budget and set of preferences, then authorize it to complete the purchase autonomously.
This change collapses search, comparison, and payment into a single flow. It also shifts power away from traditional storefronts toward whichever platforms control the agent interface. For payment companies, that creates both risk and opportunity. If they fail to integrate seamlessly into agent-driven workflows, they risk being abstracted away. If they succeed, they become the trusted rails that allow intelligent systems to transact safely at scale.
Why payment networks are moving early
Companies such as Visa and Mastercard see agentic commerce as the next logical evolution of their role. Having already enabled the shift from cash to cards and from cards to online payments, they now aim to underpin transactions initiated by non-human actors. The core logic is defensive as much as innovative: if AI agents are going to spend money, the existing payment infrastructure must be ready to recognize, authenticate, and manage them.
Both networks have spent the past year building protocols that allow AI agents to act with delegated authority. These systems are designed to verify that a bot is acting on behalf of a real, consenting user and within clearly defined limits. Without such guardrails, agent-driven commerce would be vulnerable to fraud, abuse, and costly disputes.
In practice, agentic commerce involves three layers working together. The first is the consumer interface, typically a conversational AI system where the user expresses intent. The second is the decision layer, where the agent searches, filters, and evaluates options. The third is the transaction layer, where payment credentials are invoked and funds are transferred.
Payment companies focus almost entirely on this final layer, but it must integrate tightly with the others. A flight-booking agent, for example, must not only find the cheapest acceptable route but also confirm that it is authorized to spend up to a specific amount using a specific payment method. Once those conditions are met, the payment needs to clear instantly, without requiring additional human intervention.
This capability opens new use cases. Agents can monitor prices continuously and execute purchases when conditions are met, even while the user is offline. They can manage subscriptions, optimize travel itineraries, or reorder essentials automatically. For consumers, the appeal lies in time savings and reduced cognitive load. For payment firms, it lies in preserving transaction volume as behavior changes.
Early pilots and ecosystem alignment
Both Visa and Mastercard have already run pilot programs with selected merchants and users, testing how agent-initiated transactions behave in real-world environments. These pilots focus heavily on authentication, ensuring that authorized agents can be distinguished from malicious bots attempting to exploit automated systems.
Agentic transactions are expected to occur across widely used AI platforms rather than standalone payment apps. Systems such as OpenAI’s ChatGPT and Google’s Gemini are emerging as primary interfaces for consumer agents, while merchants and banks are experimenting with proprietary agents tailored to their ecosystems.
Payment companies are working directly with these AI providers to ensure compatibility. The goal is to make payment execution invisible to the end user while maintaining the same—or higher—levels of security and dispute protection that exist today.
For merchants, agentic commerce is more disruptive than traditional e-commerce. AI agents are indifferent to branding, page design, or emotional appeals unless explicitly instructed otherwise. They prioritize price, availability, delivery speed, and reliability. This threatens established marketing strategies built around impulse purchases and visual merchandising.
Large retailers are responding defensively. Amazon has begun testing its own internal agentic tools while restricting external agents from freely accessing its platform. The strategy reflects a desire to retain control over customer relationships and data, even as consumer behavior shifts toward AI-mediated purchasing.
Smaller merchants face a different challenge. Without scale or proprietary agents, they risk being filtered out by algorithms that favor efficiency and price competitiveness. Payment firms expect merchants to adapt by offering machine-readable incentives, loyalty structures designed for agents, and real-time pricing transparency.
The security and liability problem
The most complex issues in agentic commerce emerge after something goes wrong. Traditional payment disputes involve four parties: the consumer, the issuing bank, the acquiring bank, and the merchant. Agentic commerce inserts a fifth: the AI platform or agent provider. Determining responsibility when an agent books the wrong date, purchases the wrong item, or exceeds its mandate is far from straightforward.
Payment networks are responding by building cryptographic authentication systems that bind agents to specific users and permissions. Visa’s collaboration with Cloudflare reflects this focus, creating verifiable credentials that prove an agent is authorized to act and define what it is allowed to do.
Beyond authentication, banks are being given richer “payment signals” that include contextual data about agent behavior. This allows risk engines to distinguish legitimate automation from suspicious activity. Still, the assumption within the industry is that errors will occur and that dispute frameworks must be redesigned accordingly.
A shift in competitive power
Agentic commerce also alters competitive dynamics within payments themselves. If AI platforms were to embed proprietary payment systems deeply into their agents, traditional networks could be sidelined. This risk explains why payment giants are moving early to make themselves indispensable to agent workflows, rather than reactive intermediaries.
Partnerships between AI platforms and alternative payment providers highlight this tension. PayPal’s collaboration with Perplexity to enable agent-based shopping is one example of how the ecosystem is fragmenting. For Visa and Mastercard, scale and global acceptance remain their strongest advantages—but only if they can translate those advantages into the agentic era.
Supporters argue that agentic commerce will ultimately benefit consumers by lowering search costs and increasing price transparency. When agents can instantly compare thousands of offers, inefficient pricing becomes harder to sustain. Over time, this could compress margins in some sectors while rewarding merchants that optimize logistics and reliability.
At the same time, consumer behavior may become more passive. Instead of browsing and choosing, users will increasingly set parameters and trust agents to act. That trust places enormous responsibility on both AI developers and payment networks to ensure outcomes align with user intent.
An unavoidable transition
Within the payments industry, there is little debate about whether agentic commerce will happen. The remaining uncertainty is timing and scale. Executives increasingly suggest that meaningful adoption is a matter of months rather than years, driven by rapid improvements in AI capability and consumer comfort with delegation.
For payment giants, preparing for this world is not optional. It requires rethinking authentication, redesigning liability frameworks, and negotiating new power balances with AI platforms and merchants. The result is a payments landscape where intelligence, not just speed, defines value—and where the ability to serve machines may matter as much as the ability to serve people.
(Source:www.cnbc.com)
Executives across the sector increasingly describe this transition as inevitable. As consumers grow comfortable asking AI tools to plan trips, compare prices, and recommend products, completing the transaction within the same interface becomes the logical next step. The challenge for payment networks is ensuring that this new layer of commerce remains secure, interoperable, and economically viable.
From digital checkout to delegated decision-making
E-commerce moved payments from physical stores to screens, but it still relied on human decision-making at the point of sale. Agentic commerce goes further by delegating that decision-making itself. Instead of browsing multiple sites, a consumer can instruct an AI agent to find the best option within a defined budget and set of preferences, then authorize it to complete the purchase autonomously.
This change collapses search, comparison, and payment into a single flow. It also shifts power away from traditional storefronts toward whichever platforms control the agent interface. For payment companies, that creates both risk and opportunity. If they fail to integrate seamlessly into agent-driven workflows, they risk being abstracted away. If they succeed, they become the trusted rails that allow intelligent systems to transact safely at scale.
Why payment networks are moving early
Companies such as Visa and Mastercard see agentic commerce as the next logical evolution of their role. Having already enabled the shift from cash to cards and from cards to online payments, they now aim to underpin transactions initiated by non-human actors. The core logic is defensive as much as innovative: if AI agents are going to spend money, the existing payment infrastructure must be ready to recognize, authenticate, and manage them.
Both networks have spent the past year building protocols that allow AI agents to act with delegated authority. These systems are designed to verify that a bot is acting on behalf of a real, consenting user and within clearly defined limits. Without such guardrails, agent-driven commerce would be vulnerable to fraud, abuse, and costly disputes.
In practice, agentic commerce involves three layers working together. The first is the consumer interface, typically a conversational AI system where the user expresses intent. The second is the decision layer, where the agent searches, filters, and evaluates options. The third is the transaction layer, where payment credentials are invoked and funds are transferred.
Payment companies focus almost entirely on this final layer, but it must integrate tightly with the others. A flight-booking agent, for example, must not only find the cheapest acceptable route but also confirm that it is authorized to spend up to a specific amount using a specific payment method. Once those conditions are met, the payment needs to clear instantly, without requiring additional human intervention.
This capability opens new use cases. Agents can monitor prices continuously and execute purchases when conditions are met, even while the user is offline. They can manage subscriptions, optimize travel itineraries, or reorder essentials automatically. For consumers, the appeal lies in time savings and reduced cognitive load. For payment firms, it lies in preserving transaction volume as behavior changes.
Early pilots and ecosystem alignment
Both Visa and Mastercard have already run pilot programs with selected merchants and users, testing how agent-initiated transactions behave in real-world environments. These pilots focus heavily on authentication, ensuring that authorized agents can be distinguished from malicious bots attempting to exploit automated systems.
Agentic transactions are expected to occur across widely used AI platforms rather than standalone payment apps. Systems such as OpenAI’s ChatGPT and Google’s Gemini are emerging as primary interfaces for consumer agents, while merchants and banks are experimenting with proprietary agents tailored to their ecosystems.
Payment companies are working directly with these AI providers to ensure compatibility. The goal is to make payment execution invisible to the end user while maintaining the same—or higher—levels of security and dispute protection that exist today.
For merchants, agentic commerce is more disruptive than traditional e-commerce. AI agents are indifferent to branding, page design, or emotional appeals unless explicitly instructed otherwise. They prioritize price, availability, delivery speed, and reliability. This threatens established marketing strategies built around impulse purchases and visual merchandising.
Large retailers are responding defensively. Amazon has begun testing its own internal agentic tools while restricting external agents from freely accessing its platform. The strategy reflects a desire to retain control over customer relationships and data, even as consumer behavior shifts toward AI-mediated purchasing.
Smaller merchants face a different challenge. Without scale or proprietary agents, they risk being filtered out by algorithms that favor efficiency and price competitiveness. Payment firms expect merchants to adapt by offering machine-readable incentives, loyalty structures designed for agents, and real-time pricing transparency.
The security and liability problem
The most complex issues in agentic commerce emerge after something goes wrong. Traditional payment disputes involve four parties: the consumer, the issuing bank, the acquiring bank, and the merchant. Agentic commerce inserts a fifth: the AI platform or agent provider. Determining responsibility when an agent books the wrong date, purchases the wrong item, or exceeds its mandate is far from straightforward.
Payment networks are responding by building cryptographic authentication systems that bind agents to specific users and permissions. Visa’s collaboration with Cloudflare reflects this focus, creating verifiable credentials that prove an agent is authorized to act and define what it is allowed to do.
Beyond authentication, banks are being given richer “payment signals” that include contextual data about agent behavior. This allows risk engines to distinguish legitimate automation from suspicious activity. Still, the assumption within the industry is that errors will occur and that dispute frameworks must be redesigned accordingly.
A shift in competitive power
Agentic commerce also alters competitive dynamics within payments themselves. If AI platforms were to embed proprietary payment systems deeply into their agents, traditional networks could be sidelined. This risk explains why payment giants are moving early to make themselves indispensable to agent workflows, rather than reactive intermediaries.
Partnerships between AI platforms and alternative payment providers highlight this tension. PayPal’s collaboration with Perplexity to enable agent-based shopping is one example of how the ecosystem is fragmenting. For Visa and Mastercard, scale and global acceptance remain their strongest advantages—but only if they can translate those advantages into the agentic era.
Supporters argue that agentic commerce will ultimately benefit consumers by lowering search costs and increasing price transparency. When agents can instantly compare thousands of offers, inefficient pricing becomes harder to sustain. Over time, this could compress margins in some sectors while rewarding merchants that optimize logistics and reliability.
At the same time, consumer behavior may become more passive. Instead of browsing and choosing, users will increasingly set parameters and trust agents to act. That trust places enormous responsibility on both AI developers and payment networks to ensure outcomes align with user intent.
An unavoidable transition
Within the payments industry, there is little debate about whether agentic commerce will happen. The remaining uncertainty is timing and scale. Executives increasingly suggest that meaningful adoption is a matter of months rather than years, driven by rapid improvements in AI capability and consumer comfort with delegation.
For payment giants, preparing for this world is not optional. It requires rethinking authentication, redesigning liability frameworks, and negotiating new power balances with AI platforms and merchants. The result is a payments landscape where intelligence, not just speed, defines value—and where the ability to serve machines may matter as much as the ability to serve people.
(Source:www.cnbc.com)
Homepage
Send to a friend
Printable version
Share
