Agent commerce: how AI agents will start buying for their customers.

In March 2026, an artificial intelligence agent searched for, selected, and paid for a product on its own, within limits defined by a person and without them opening a single website. The operation was conducted by Banco do Brasil and Visa , lasted about a minute, and was recorded as the country's first agentic commerce transaction, according to Visa itself.

This wasn't an isolated experiment, but the first sign of a shift that is already driving the biggest players in technology and payments worldwide. In agentic commerce, the person who researches, compares, and finalizes the purchase is no longer always the individual, but increasingly an AI agent acting on their behalf . For digital retail, this reorganizes the most important business question: it's no longer just " how to convince the customer ," but rather " how to be found and chosen by those who buy for them ."

Agency trade: what is already a fact

• Agent commerce is a model in which AI agents autonomously research, compare, negotiate, and complete purchases within parameters defined by the consumer. (product, price ceiling, deadline). The person delegates the execution; the agent acts.
• It already influences sales on a large scale.according to year-end data from 2025 SalesforceAI influenced 20% of Sales global online data during that period, equivalent to US $ 262 billion.
• The potential is structural, not marginal: a McKinsey It projects that AI agents can orchestrate US$3 to US$5 trillion in global retail spending by 2030.
• Brazil is ahead in its readiness to use: by the study Visa Agentic CommerceBrazilian consumers are among the most willing in the world to delegate purchasing decisions to agents.
• Visibility changes its logic.The agent is not sensitive to page or ad design; it reads structured data. Those who don't display their catalog, prices, and inventory in a machine-readable format tend to disappear from this marketplace.
• It doesn't require a new website.The adjustment involves data and integration, complete product information, consistent pricing and inventory, and accessible APIs.

What is agent trading in practice?

Agent commerce is a form of buying and selling in which AI agents act on behalf of the consumer (or the company) to research, compare, negotiate, and complete a transaction , often without human intervention at each step.

The difference compared to traditional e-commerce lies in the work involved. In the current model, the person does everything: searches, opens tabs, reads reviews, compares prices, applies coupons, and finalizes the purchase. In agent commerce, the person simply states their intention, " I want an ergonomic chair up to R$900 with delivery by Friday, " and the agent executes: consults several stores simultaneously, filters by price and delivery time, checks stock, applies discounts, and completes the order with a previously authorized payment method.

It's important to distinguish between two concepts that are often confused. AI-powered e-commerce uses technology to improve the customer experience: personalized recommendations, intelligent search, and automated customer service. Agent-based commerce goes further: the agent takes over part of the decision-making and execution in place of the person. One suggests; the other acts.

Why has agent trading become a trend now?

Three movements occurred almost simultaneously and unlocked the model:

The first is behavioral. A global survey by Riskified with more than 5.000 consumers (including Brazil) showed that almost three out of four shoppers (73%) already use AI at some point in the buying journey, mainly to get product ideas (45%), summarize reviews (37%), and compare prices (32%). The habit of " asking before buying " is already established.

The second is payment infrastructure. Banco do Brasil's transaction with Visa showed that the path exists: in April 2026, Visa launched the Visa Agentic Ready program in Brazil , a controlled environment for partners to test and validate purchases initiated by agents, with tokenization and biometric authentication linking each transaction to a real person.

The third is protocols. AI platforms have begun to standardize how agents and stores communicate, transforming isolated initiatives into a market standard. It is this set of open standards that allows an agent to read the catalog, check the price in real time, and complete the purchase without depending on the website layout.

How the buyer's journey changes with agentic commerce

The impact becomes clearer when comparing the purchase journey in the two models.

There's an important business effect embedded in this. When traffic arrives via AI, it tends to be more qualified because it has already gone through a filtering and comparison stage. Salesforce observed that traffic referred by AI converts at a rate about eight times higher than that of social media. The interpretation isn't "fewer people on the site," but rather "those who arrive are more decisive." This shifts the focus from the number of visits metric to the quality of the data that informs the agent's decision.

How to prepare your store for agency commerce.

Getting ready isn't an isolated IT project, nor does it require rebuilding the website from scratch. The agent buys from the store they can read; the work is on four fronts.

1- Machine-readable product data

The agent doesn't find the website "pretty," it reads data. Catalogs with free-text descriptions, images without metadata, and incomplete listings become invisible. The starting point is to audit the completeness of the catalog and structure attributes, price, and availability into a standard that AI systems can process.

2- Open and up-to-date APIs

To be included in the agent's comparisons, the operation needs to display real-time price, stock, delivery time, exchange policy, and payment terms through accessible APIs. Without this, the product may exist, but it doesn't participate in the decision-making process.

3- A unique and consistent foundation

It's pointless for the data to be legible if the price displayed in the store differs from the stock price, or if one channel shows availability that another doesn't confirm. Registration, pricing, stock, and tax information must all originate from the same source. This consistency is what sustains the agent's trust in your store.

4- Your own sales and customer service agents

Being discovered by third-party agents is half the game. The other half is having your own agents who accompany the customer, respond in natural language, and complete transactions within conversational channels , without depending on a website visit. This is exactly the terrain of the VTEX CX Platform , built to operate natively on catalog, checkout, and order management , allowing the agent not only to suggest but to execute the sale, exchange, or cart recovery from start to finish.

Is agency trading already a reality in Brazil?

Yes, it's still in the early stages, but it's underway. The country's first agent transaction has already been executed, the payment gateway already has a certified environment, and Brazilian consumers are among the most open to delegating purchases . For digital retail, the cost of waiting is not neutral: while the database and integrations are not ready, the store simply doesn't appear to those who already buy through agents.

The good news is that preparation is orderly and predictable when conducted with those who know the terrain . Structuring a catalog, exposing APIs, and adopting a platform prepared for this new phase is a natural step in evolution and comes sooner for those who start now.

Prepare your operation for the next phase of digital retail.

Agent commerce is an ongoing movement, and the difference between being in or out of it starts with the store's technological foundation. Migrating to or implementing a platform prepared for this reality, such as VTEX , is a growth step that yields greater results when planned in advance and with a partner who has already made this journey many times.

For over 13 years, e-Plus has been implementing, migrating, and evolving e-commerce operations on market-leading platforms. If you want to understand how to make your catalog readable by AI agents and structure your store for this new phase, talk to the e-Plus team of experts and discuss the best path for your operation. To further explore this topic, also check out other blog content about AI in e-commerce.

Frequently Asked Questions – Agency Trade