AI and Retail
A potential market of 310,000 million euros in Europe. And yet, 94% of consumers want to approve before the agent takes action. Data from 8,400 European consumers reveals what users are willing to delegate — and where they don't cross the line.
The agent who buys for you: what the European consumer is willing to delegate and what he is not.
There is a moment when AI stops being an assistant and becomes an agent. The difference is not technical — it is control. An assistant helps you decide. An agent decides and acts for you.
That moment is coming to European trade. And the data shows that consumers have a very clear opinion about when they are willing to cross that line.
A Sopra Steria study conducted in February 2026 with 8,400 consumers in eight European countries — including Spain — reveals that more than half of Europeans (55%) already know the concept of agentic commerce: AI agents capable of searching, comparing, negotiating and executing purchases autonomously on behalf of the user. In Spain, familiarity with the concept is above the European average.
The potential market in Europe is estimated at 310 billion euros in the next ten years. It's not an emerging trend — it's a transformation already underway in the United States, where OpenAI, Google and Amazon have deployed their first purchasing agents.
What the Spanish consumer is willing to delegate
The study data reveals a clear logic in which categories Europeans are willing to trust an agent: technical or routine purchases come before personal or sensitive ones.
In Spain, 50% of consumers consider the possibility of delegating electronics purchases to an AI agent. 39% would do it with energy. But only 18% would do so with food — where personal preference matters — and 9% with health or financial services, where the level of personal involvement is maximum.
The conclusion is not that agentic trade will take time to reach Spain. The thing is that it will reach some categories first and then others, and the design of the experience has to respect that logic.
A travel platform that offers agentic delegation is on a different terrain than an insurer that wants an agent to manage policy renewals without explicit user confirmation. Not because the technology is different — but because the user's relationship with those categories is different.
94% want to approve before the agent acts
The most revealing data from the study is also the clearest: 94% of European consumers want to approve each transaction before the agent executes it.
That's not a nuance. It is a starting condition for any agentic experience design. The user can delegate the search, compare and select — but wants the confirmation button. He wants to know what is going to happen before it happens.
The main concerns that the study identifies confirm this pattern: unwanted purchases or errors (49% cite this as their first concern), loss of control over spending (48%), theft of bank details (47%), manipulation by commercial interests (45%).
And the guarantees that consumers demand before delegating are equally explicit: transparency of the agent's selection criteria (75%), ease of cancellation and returns (64%), protection of personal data (55%), absence of hidden commissions (54%).
These are not usability preferences. They are conditions of trust. And without trust, agentic commerce doesn't work — no matter how good the technology is.
The opportunity for Spanish banks
Perhaps the most surprising finding from the study is who wins the trust battle when consumers are asked who should develop their personal buying agent.
Banks lead with 27%. The big American technology companies remain in the 10%. The large European retailers, at 6%.
That's a reversal of the order anyone would have predicted five years ago. And it has enormous strategic implications for Spanish financial entities.
Banks have something that neither OpenAI nor Amazon have with their European users: a relationship of trust built over decades, complete transactional data on customer purchasing behavior, and an already consolidated payments infrastructure. If they design the agentic experience well — with transparency, with real user control, with explainable selection criteria — they have a real competitive advantage over the American technology giants.
But that advantage is not automatic. Requires design. It requires that the agent experience be thought of from the perspective of the user, not from the perspective of maximizing the transaction. A bank that builds a purchasing agent that the user feels works for them — and not for the interests of bank or its business partners — has a unique position in this market.
The European regulatory framework as an advantage, not as a brake
The European AI Regulation, GDPR, and Payment Services Directive (PSD2) create a more demanding framework for agentic commerce in Europe than exists in the United States.
That framework can be read as an obstacle. Or it can be read as a competitive advantage for European actors who comply well: if 75% of consumers demand transparency about the agent's criteria, a regulation that forces this transparency does not stop business — it enables it.
Trust in European agentic commerce will be built on transparency, reversibility and user protection. The actors who design from these principles — not to comply with the regulatory minimum, but so that the user feels real — are the ones who will win the adoption battle.