The discussion surrounding artificial intelligence in retail has gone through several stages of development in recent years. Following search engine optimization (SEO), personalized recommendations, and generative AI, the next stage of evolution is now emerging: Agentic Commerce.

In Agentic Commerce, consumers and businesses use autonomous AI agents that not only provide information but also actively take on tasks, make decisions, and—in the future—may even carry out entire transactions on their own. For retailers and brand manufacturers, this represents a fundamental transformation of the customer journey. It is no longer just people who search for, compare, and purchase products—AI agents will increasingly take on these tasks. Agentic Commerce is therefore considered the next stage of development in digital commerce.

A joint study by Arvato Systems and ECC KÖLN, scheduled for publication in September 2026, focuses precisely on this development. Initial findings were already presented at this year’s K5 Future Retail Conference as part of the masterclass “Agentic Commerce: Between Vision and Reality.” The results show that many companies expect autonomous procurement and sales agents to bring about profound changes to existing sales structures.

But what does Agentic Commerce actually look like? What use cases are already relevant for retailers and brands today? Three use cases illustrate the potential of agent-based shopping experiences.
 

Agentic Commerce refers to digital commerce processes in which AI agents handle one or more steps of the purchasing process. These include product search, comparison of alternatives, evaluation of offers, purchase decision, and—in the future—also ordering and payment. The agent acts according to rules, preferences, and permissions granted by the user.

This has far-reaching consequences for retailers and brands:

• Product data must be machine-readable and up-to-date.
• Prices and inventory must be available in real time.
• E-commerce systems require APIs and open interfaces.
• Trust is becoming a competitive factor.
• Transactions must also be possible outside the traditional store.

In the future, the key question will no longer be, “Is my store user-friendly?” but rather, “Can an AI agent understand, evaluate, and make use of my offerings?”

Imagine you want to buy a specific pair of sneakers. Instead of comparing prices yourself regularly, set up a task for your AI agent: "Find me this model in size 42. Buy it automatically as soon as the price drops below 120 euros."
 

The agent then monitors various stores, marketplaces, and offers. It evaluates prices, delivery times, and availability, and places the order as soon as the defined conditions are met. What many consumers still do manually today can be automated with Agentic Commerce. The AI agent becomes a digital bargain hunter.

For retailers, this represents a paradigm shift. In the future, products will no longer compete solely for people’s attention, but also for the preferences of agents. Criteria such as availability, price, delivery speed, and return policies are becoming increasingly important. At the same time, the influence of traditional marketing tools is declining in the actual purchasing situation.

In the commodities sector in particular, this could significantly intensify competition. When agents objectively compare offers, product data quality, transparency, and operational excellence become more important than mere visibility or advertising budgets. Study results suggest that companies expect hard criteria such as price, availability, and sustainability to carry greater weight in the future.

At the same time, this presents an opportunity for brands to build new signals of trust. Certificates, reviews, sustainability information, or the Digital Product Passport could become important reference points for AI agents in the future. Those who provide this information in a structured manner increase their chances of being selected by agents.
 

Few things frustrate customers more than out-of-stock items. In traditional e-commerce, the customer journey often ends right at this point. Agentic Commerce introduces a completely new approach. Instead of leaving the page, the customer activates their agent: "Notify me as soon as the product is back in stock—and order it automatically."

The agent then monitors availability and initiates the purchase immediately upon receipt of the goods. The interaction evolves from a one-time visit into an ongoing process.

In the K5 Masterclass, this very scenario was described as one of the most exciting use cases. The core idea is: In the past, “sold out” meant lost revenue. In the future, “out of stock” could be the moment when the customer activates their agent.

This use case seems particularly relevant in industries with limited inventory or high demand, such as fashion, collectibles, or consumer electronics.

However, this requires reliable inventory data and real-time interfaces. Agents can only act effectively if availability information is provided accurately and immediately. Incorrect product data, on the other hand, would directly lead to poor decisions and abandoned purchases. This is precisely why data quality is becoming a key requirement for agentic commerce.

Perhaps the most far-reaching step in agentic commerce is the autonomous execution of transactions. The process is very simple:

1. Users define purchase rules.
2. The agent searches for suitable products.
3. The agent selects an offer.
4. Payment is made within the agent environment.
5. The order is being processed.

To ensure this works reliably, agents are not issued traditional credit cards. Instead, controlled permissions and tokens are used. The user defines exactly what may be purchased, the price range within which the agent may make purchases, and the conditions under which a transaction is permitted. All activity remains traceable and documented.

This use case is particularly relevant for retailers and brands because it eliminates a major source of friction in digital commerce: the switch between product search and checkout. The purchase is completed right where the agent is already active—regardless of whether that’s on an AI platform, a messaging app, a mobile app, or, in the future, in an entirely new digital context. The implications are far-reaching:

• E-commerce is expanding beyond the traditional online store.
• New distribution channels are emerging.
• Payment becomes part of agent-based processes.
• Merchants must ensure they can process transactions even outside their own stores.

Agentic Payment could thus become a key component of the future commerce architecture.

As part of a joint pilot project involving Arvato Systems, the payment provider Nuvei, its partner Visa, and the fashion brand Kings and Priests, this very scenario has already been implemented in practice: For the first time, an AI agent completed an entire transaction within an agent environment—without being redirected to a traditional checkout process. Read more about this in our press release.

Agentic Commerce is no longer just a theoretical vision of the future. The first proofs of concept are already demonstrating how AI agents can search for products, monitor availability, and even complete purchases on their own. Retailers and brands are thus facing a fundamental transformation of their digital sales channels. The three use cases presented here clearly show where this is headed:

• The AI agent becomes a digital bargain hunter.
• Out-of-stock situations become new sales opportunities.
• Autonomous payments open up new sales channels.

Those who start now to prepare their data, processes, and systems for this development will lay the groundwork for the next stage in the evolution of digital commerce.