Agentic Commerce: From Chat to Commerce Infrastructure
24 July 2026
By Paul Adams – Chief Growth Officer
Agentic commerce is moving beyond clever chatbots. The real shift is that AI is starting to behave like a shopping layer that can help people compare products, make choices, and complete purchases. That is a much bigger change than adding a friendly assistant to a website. It means brands need to think less about “How do we make the chatbot better?” and more about “How do we make our products easy for AI to find, understand, trust, and buy?” That is the part many brands still miss. A chatbot can answer questions. It can suggest products. It can even guide someone through a category. But agentic commerce is more than that. It is about building the commercial infrastructure that allows an AI system to do the work a shopper used to do manually: search, compare, shortlist, decide, and transact. That is why this is not really a front-end trend. It is a whole-commerce shift.
Over the last three months, the market has become more explicit about what agentic commerce actually requires. The shiny “chatbot layer” is no longer the story. The real focus has moved to the systems underneath: product data, availability, pricing, payment trust, checkout, and fulfilment. In plain terms, the industry has realised that the value is not in making AI talk well. It is in making AI able to act well.
That is a meaningful change. A brand might use ChatGPT, Claude, Gemini, Perplexity, Amazon Rufus, or a custom assistant to help a shopper narrow choices. But if the product feed is messy, if stock is wrong, if delivery logic is unclear, or if checkout creates friction, the agent cannot complete the job properly. The consumer may never even see the brand if the data is not usable.
We are also seeing more real examples of this shift. AI shopping agents are already being used in practical scenarios such as travel booking, replenishment, and category shopping where the buyer has a clear goal and a few constraints. In other words, “find me the best option under this budget” is becoming more common than “have a chat with our assistant”. That distinction matters. One is a conversation. The other is a commercial workflow.
Seen through that lens, the last six months look different. A lot of what appeared exciting was really early experimentation. Useful, yes. But still early. The deeper progress was happening in the background, where brands and platforms were making their commerce stacks easier for machines to work with.
That means some of the noise now feels less impressive. A chatbot that sounds polished but cannot support a serious buying journey is not a long-term differentiator. It may help with basic browsing. It may create a decent demo. But it is not where the market is heading. The brands that are building advantage are the ones thinking about machine-readable commerce, not just machine-friendly conversation.
A better example is this: if a shopper asks an AI, “What is the best running shoe for long road runs under £120, available in my size and delivered by Friday?” the agent needs structured data, accurate stock, and reliable delivery information to answer properly. If your systems cannot support that, your brand is not really in the race. That is the new standard.
This also changes how we should think about traffic. AI-led discovery is starting to shape how shoppers find products, compare options, and move towards purchase. That means classic search, paid media, and homepage merchandising still matter, but they are no longer enough on their own. Visibility is becoming tied to how well a brand’s commercial data is structured and maintained.
The US and Europe are moving differently, and that difference is important. The US is moving faster because the biggest AI and commerce players are more aligned, and brands are more willing to test quickly. Europe is moving more cautiously. It is more fragmented, more regulated, and more complex across markets, languages, and payment environments. That does not mean Europe is behind in any simple sense. It means it is solving a harder problem first. In the US, the emphasis is on speed and experimentation. In Europe, the emphasis is on trust, consent, and control. A shopper’s journey is more likely to break at the point of payment in Europe because the rules around authorisation, liability, and consumer protection are stricter. That slows adoption, but it also forces better discipline. For brands, the implication is straightforward. The US is the place to move fast and learn fast. Europe is the place to build carefully and locally. The same playbook will not work in both regions. At Luzern, we hear a lot from clients about how they are building a ‘Global Playbook’ for Agentic Commerce. These playbooks are normally developing in the US with a US perspectives and therefore can’t be executed in Europe. The smart brands are developing local strategies and infrastructure and reapplying country by country.
The next stage of agentic commerce will be about who gets selected, not just who gets seen. AI agents will increasingly act like a filter between the shopper and the brand. They will compare options, narrow the field, and shorten the buying journey. That is already changing the balance of power.
For brands, the practical response to the changing trend is clear. Get the basics right. Make your product data clean. Keep stock and pricing accurate. Remove friction from checkout. Strengthen trust signals. Build systems that are easy for AI to work with. This is where the battle is being won in categories. The brands that succeed will not be the ones that are seen. They will be the ones with the most usable commerce infrastructure. That is the real shift. Not chatbot theatre. Not a shiny layer. A new operating model for online selling.
Luzern eCommerce are specialists in Agentic Commerce. We are currently working with Beta Partners to test our technology and service before a full market launch in the autumn. You can find out more about what we offer at https://www.luzern.co/channels/agentic-commerce/ and use our Contact Us form to book an appointment.