Innovations related to the development of mobile payments have simplified the shopping process, but AI is pushing the boundaries of e-commerce even further.
Over the past two decades, e-commerce has focused primarily on expanding the availability of payment methods, speeding up the purchase process and increasing conversions. Subsequent innovations related to the rise of mobile payments, such as one-click payments or digital wallets, simplified the purchase process but still assumed an active role for the user.
Today, however, we are starting to see a clear shift in purchasing patterns: from e-commerce driven by consumers to agentic purchases, in which AI systems acting on the user’s behalf take over a significant part of purchasing decisions.
Before autonomous AI agents definitively become the main drivers of e-commerce, the market is reaching (and already has reached) maturity in subscriptions. Subscriptions are now the natural transitional phase between classic e-commerce and automated agentic shopping. The user does not have to initiate every purchase: the decision is automated in advance, while the user keeps control over the agreed terms.
The dynamic growth of subscriptions is already visible not only in digital services but also in the sale of physical products. New players are also entering the race for subscribers. One example is payments using saved cards in Apple Pay and Google Pay. Everything suggests that 2026 will be a period of intense expansion of subscriptions, which will prepare the ground for further phases of purchase automation.
Agentic purchasing is an online commerce model in which AI systems can operate at various levels of autonomy:
Agentic commerce has the potential to become an important direction in the development of online commerce, but current market initiatives are still at the pilot project stage. Solutions developed by global players in the fintech market do not create a new payment infrastructure but build on existing technology systems, in particular tokenization, card standards and authentication mechanisms.
Visa is developing the concept of intelligent commerce, Google is working on the Agent Payments Protocol and Mastercard is testing approaches referred to as agentic payments, using network token extensions and existing authorization models to securely process payments initiated by AI systems.
The first experiments with this type of solution can be expected here in the Czech Republic as early as this year, although the question of liability for the actions of AI agents in financial transactions still remains a major challenge.
Online payments are becoming a key element of agentic commerce. Secure, predictable and fully trustworthy mechanisms are a necessary prerequisite for users to hand over part of the control to autonomous systems.
In the early phase of development, artificial intelligence served mainly as an “invisible guardian”. It was used for fraud detection, anomaly identification and risk assessment in real time. Its role was reactive and security-focused. An example of such a solution is the Antifraud 24/7 model from PayU.
Now AI is increasingly becoming an active moderator of purchases, influencing the course of the purchase process and the decisions made at the payment completion stage. Alongside payment security, attention must also extend to designing a functional shopping experience.
Artificial intelligence is now beginning to define the nature of online payments. They are ceasing to be a neutral mechanism for transferring money and are gradually becoming an interactive environment that adapts to the user. In the world of agentic shopping, payments will no longer be the final stage of the purchase process but will be an integral part of it.
The future of online payments is not only about speed or reducing transaction delays. Its essence will be designing payments that are intuitive, secure and at the same time personalized, enabling users to trust AI systems as full participants in the purchase process.