Agentic commerce has begun: how retailers can prepare to sell to AIs

By Alex Marques, commercial director at Data System

Retail has always improved its ability to understand people, studying their consumption habits, purchasing behavior, digital journeys and ways to make the customer experience more efficient. In addition to investing in storefronts, campaigns, marketplaces, omnichannel and artificial intelligence to personalize offers. Now, a new transformation is beginning to gain ground and could change the rules of the game again. For the first time, the person making a purchase may not be a person, but rather an AI agent.

Agentic Commerce has already become a reality. In early 2026, Google, Shopify, Target and other major retail players launched the Universal Commerce Protocol (UCP), an open standard created to allow artificial intelligence agents to search, compare products and complete purchases seamlessly across different platforms. In Brazil, Banco do Brasil and Visa also carried out the country's first fully agentic transaction, in which an AI agent searched, selected and completed a purchase on behalf of the consumer within previously defined parameters.

In this new model, the user simply informs their objective, such as buying a notebook with certain specifications, renewing pantry products or finding the best airline ticket within a budget and the AI ​​researches suppliers, compares alternatives, checks stock availability, analyzes reviews, calculates delivery times and completes the purchase according to the preferences established by the consumer.

The purchasing decision becomes data-driven

This change profoundly alters the way in which retail competes for consumer attention, because until now much of the investment has been focused on winning over people. Intuitive layouts, good images and creative campaigns remain fundamental, but AI agents do not buy on impulse nor are they influenced by a well-produced showcase, they make decisions based on objective criteria and the quality of available information.

In practice, this means that the new consumer will be convinced by data. An intelligent agent compares technical specifications, availability, price history, reviews, shipping cost, delivery time and seller reputation. The more complete, structured and reliable this data is, the greater the chance of a product appearing among the recommendations. If the information is incomplete or inconsistent, the consequence may be that that product ceases to exist for artificial intelligence.

It is precisely at this point that the sector's main challenge lies: even with years of digitalization, product registration is usually treated as an operational rather than a strategic activity. Superficial descriptions, attributes filled out differently between channels, incorrect categories, and outdated information are still common. Today, this reduces efficiency and conversion. Tomorrow could mean invisibility for the AI ​​agents responsible for purchasing decisions.

Being found will no longer depend solely on traditional search engines, companies will need to ensure that their data is understood by AIs capable of comparing thousands of options in seconds and automatically selecting the best alternatives for each consumer.

ERP and product registration gain prominence

In this scenario, management systems become strategic assets, ERP is no longer just an administrative platform to concentrate reliable information about products, stock, prices, availability and logistics. The quality of this data directly influences a retailer's ability to participate in the recommendations made by intelligent agents.

This transformation also requires a cultural change, as quality data is not just the responsibility of the technology area. Marketing, purchasing, commercial, logistics and operations participate daily in building the information that feeds the systems. When this data is treated as a strategic asset, the entire company becomes more prepared to compete in an increasingly automated environment.

The future of retail starts now

Another important aspect involves payment methods, with this movement beginning to take shape with initiatives such as the Visa Agentic Ready program, which prepares the payments ecosystem to allow AI agents to make purchases on behalf of consumers securely, within previously authorized limits and rules. As this model advances, banks, payment platforms and retailers will need to evolve their systems to guarantee authentication, traceability and security in all transactions, making it possible for artificial intelligence agents to act autonomously without giving up trust and control on the part of the consumer.

At the same time, challenges related to trust arise, such as ensuring that an agent is making decisions aligned with the consumer's interests? How to avoid manipulations in recommendations? These discussions will follow the evolution of this new consumption model.

People will continue to buy directly in many situations, but a growing portion of recurring purchases based on objective criteria tend to be automated. This means that retailers will need to learn to talk to two audiences at the same time: consumers and artificial intelligence.

In the era of agentic commerce, well-structured data becomes the language that will allow retailers to be found, understood and chosen by the next consumer and this consumer may not be human.