Nearly 80 Percent of Polish Companies Open to AI-Driven Purchasing, Adyen Study Finds

Nearly 80 percent of companies operating in Poland say they are open to agentic commerce, a model of trade in which autonomous AI agents make purchasing decisions on a customer's behalf. That is according to the "2026 EMEA Research" report published on July 15, 2026 by Dutch payments company Adyen, which also shows that this declared openness rarely translates into actual deployment for now.

Agentic commerce is a concept in which users no longer browse online stores themselves but instead hand the shopping task to an AI agent, for instance an assistant built into a browser or banking app. The agent compares offers, negotiates terms, selects a vendor and completes the transaction, often without a human directly involved in picking the specific product. Adyen, which processes payments for major e-commerce platforms, is studying the trend from the perspective of both merchants and consumers.

Openness without deployment

The study reveals a clear gap between what companies say and what they actually do. While nearly four out of five firms say they are ready for agentic commerce, only just over a quarter treat it as a strategic priority for the coming year, and a quarter plan real investment. For most businesses, in other words, the topic remains at the stage of watching the market rather than rolling out concrete solutions.

The report's authors attribute this caution to a mix of factors. On one hand, companies see the potential of a new sales channel; on the other, they fear losing control over the customer relationship once an algorithm, rather than the brand, becomes the intermediary in the transaction.

Brand and trust concerns

The biggest worry, cited by 38 percent of surveyed companies, is the weakening of direct customer relationships and declining brand loyalty when AI, rather than the consumer, decides which product to buy. Companies are also grappling with how to build recognition when it is not a person browsing the offer, but a language model comparing product data.

The study also shows that the trust problem cuts both ways. About a third of the consumers surveyed doubt whether an AI will actually pick the best offer for them, rather than, say, a product promoted by the agent's provider. The question of transparency in purchase-recommending algorithms also remains open for regulators in the European Union.

In discussions of agentic commerce, we often focus on whether consumers will be willing to hand AI part of their purchasing decisions. From a commerce perspective, though, it's just as important how the role of the brand itself will change. If AI becomes the main point of contact with an offer, companies will need to learn to operate in this new customer-contact channel - Matouš Michněvič, CEE country manager at Adyen

Rising fraud risk

The report also points to the flip side of the coin: security. 24 percent of companies have recently noticed a rise in sophisticated AI-powered fraud attempts, such as fake orders generated by bots or scam attempts tailored to a specific customer. In response, 29 percent of businesses have already deployed their own AI tools to detect and curb fraud, treating artificial intelligence as both a threat and a shield.

This shows that for many companies, agentic commerce is not just a new sales channel but also a new attack surface that needs securing before it even starts generating real revenue.

What this means for Polish retail

For Polish online retailers, Adyen's findings signal that shopping agents are no longer an abstract idea but are becoming part of business planning, even if still at an early stage. Companies that are already investing in structuring their product data and improving visibility in language models could gain an edge once agentic commerce starts genuinely driving purchase traffic.

At the same time, declared openness without concrete budgets suggests that many companies are waiting for major players, such as browser makers or payment platforms, to establish technical and security standards before they invest in integrating with AI agents themselves.