Retail at a Crossroad: Competing in the Age of AI Agents

When I was at the K5 conference in Berlin earlier this summer, the buzz around AI agents in commerce was impossible to miss. At the time, I left with more questions than answers. After a few months of reflection and conversations with customers, I see the picture more clearly and the stakes feel even higher.

More and more people are planning their dates, their holidays or their grocery lists in conversations with LLMs like ChatGPT, Gemini or Perplexity. These systems don’t just respond to a query, they gather immense context. They know your habits and your interests. They can recommend better than any human clerk ever could. In a way, it feels like the old story of the US grocer predicting pregnancy from shopping behavior, but this time on steroids. From a simple query like planning a trip, the LLM already knows your history, your spending habits, the weather at your destination, and which options you’re likely to consider. It can decide whether you need shoes, a tent, or a gear upgrade. It can anticipate what you want now, as well as the kind of buyer you are. Done well, this saves enormous time and feels like liberation.

But it also makes us incredibly transparent. Corporations now hold an intimacy of data that, especially in Europe, comes with a heavy responsibility. We don’t want an almighty system that knows everything about us. Data protection laws exist for a reason, and with agentic commerce the stakes will rise even further. The EU AI Act, the Digital Markets Act and the Digital Services Act are early signals that the age of unregulated recommender systems may be coming to an end. These frameworks demand explainability, non-discrimination, and disclosure. A future where customers can ask, “Why are you recommending this?” and expect a plain-language answer is no longer far-fetched. For platforms and brands, this means that the black box logic that dominated social feeds and search rankings for the past decade won’t be acceptable anymore.

For retailers and brands, this moment feels like the early days of marketplaces. If you’re not on the platform, you risk being invisible. If you are on it, you risk losing your identity. Amazon already shows us how anonymous shopping can become. You don’t always know where your product ships from, and brand voices blur into the background. Platforms like Temu and Shein took this logic even further. They didn’t dominate by offering the best product, but by controlling visibility. Pay-to-play positioning replaced product quality as the currency of success. If AI agents follow that same path, they will become walled gardens where only sponsored placements win and retailers will once again find themselves squeezed between visibility and margin.

The impact won’t be uniform across industries. Fashion and lifestyle brands may still command loyalty, because identity and aspiration are woven into their products. A customer who insists on a certain sneaker or handbag will continue to search for that specific brand. Electronics, on the other hand, may move the other way: if the agent tells me this is the “best TV for my budget and room size,” the logo on the device might matter less than the performance specs. Grocery sits somewhere in between. We already know that substitution is common if your preferred brand of pasta isn’t in stock, the retailer nudges you to another. In an agentic model, the decision may never even reach you, the assistant could simply decide which brand lands in your basket, based on past behavior or commercial agreements. Some industries will double down on brand equity, while others risk being reduced to commodities.

Retailers in particular will feel the pressure. Their traditional promise was to be the most versatile, cheapest, easiest or most trustworthy place to buy. If the agent takes over that role, they’ll need new ways to stand out. Content will be the first battleground. For years we’ve told retailers to invest in it, now it becomes existential. LLMs thrive on organically created, information-rich content. Expert opinions, product tests, detailed specs, these are the signals that push you ahead of your competitors in an agent’s recommendations. Speed will matter just as much. If your catalogue isn’t accessible, if your checkout can’t perform under load, if your integrations are brittle, you won’t get a front row seat. This might be the right time to push data and services closer to the edge. Even beyond content and speed, retailers should be thinking about services and community. What does it mean to create experiences around the product that an agent can’t replicate? What kind of post-purchase touchpoints, loyalty systems, or human connections can extend the value of a transaction into a relationship?

Personally, I hope these agents become the guides we can trust. The idea of a system that can help me quickly, fairly, and in my interest is exciting. But I’ve also seen too many examples where greed takes over and ethics are the first thing to fall off the table. I’m cautiously optimistic. This is a great time for us as customers, but the business model behind it will decide whether it stays that way.