Your Website Has a New Job

Earlier this August, I attended CMS Connect 26 in Montréal. Many of the talks and conversations focused on AI agents, new search behavior, and the future of content management systems. Right now, it is easy to get the impression that websites and CMS platforms are becoming less important. Who still needs a website if an AI agent can research products, compare options, rule out brands, and prepare a decision? Why keep investing in a CMS if fewer people arrive directly on your pages?

I kept thinking of the phrase: The king is dead. Long live the king. Because the website is not disappearing. Its old job is. It is no longer guaranteed to be the starting point of every search. Instead, it is becoming something else: a trusted source for people and machines—and the place where an already well-prepared decision is completed

AI Is No Longer Just Helping Me Search

Recently, I asked ChatGPT whether a solid perfume would be a good fit for me. I expected a shortlist, perhaps with a few advantages, disadvantages, and links to different products. Instead, it advised me against it. Whether that recommendation was objectively correct is almost beside the point. What mattered was how I experienced it: not like a search result, but like personal advice. More like a friend who understands my requirements, weighs up the options, and is also willing to say, “This probably is not right for you.”

The AI did not help me find a website. It helped me make a decision. That is the real change. We no longer enter a few keywords and then work through ten open tabs. We have a conversation, explain what matters to us, and ask follow-up questions. The agent filters, evaluates, and rules out options. AI agents are designed to act on behalf of people or organizations, perform tasks, make decisions, and interact with other systems. As a result, a brand can fall out of consideration before a person ever visits its website.

The Traffic We Knew Is Gone

The old path was fairly predictable: Someone searched on Google, clicked a result, and started their research on your website. That path will not disappear completely, but it is becoming less important. The Forrester data presented at CMS Connect showed that consumers are increasingly using answer engines for product discovery. 

The traffic we knew is gone. It is not coming back. That sounds dramatic because, for years, we measured website success mainly through visits, sessions, and clicks. But less traffic does not automatically mean less business. Someone who reaches your website after a detailed conversation with an AI assistant is probably at a different stage in the decision. They may already understand your offer, have ruled out alternatives, and now want to answer one specific question, build confidence, or take the next step.

The visit happens later, but the intent may be clearer. That means fewer people might visit your website while a higher percentage of them convert. We should measure that rather than assume it. But it changes what we optimize for: Not every click has the same value, and not every decision generates a click anymore.

B2A Joins B2B and B2C

We are used to distinguishing between B2B and B2C. Now there is a third audience to consider: B2A—business-to-agent.

People and AI agents use content differently. People are more strongly influenced by experience, relationships, identity, and emotion. Agents work more with logic, patterns, and the context available to them. The Forrester talk described B2A as an additional strategy for B2B and B2C organizations that need to reach machine customers as well as human ones. 

That does not mean we should start writing only for machines. People still need to understand your content. Your language has to be clear, your website accessible, and the user experience good. But your content now has a second audience. An agent needs to understand what you offer, who it is for, what it costs, which conditions apply, and what evidence supports your claims. It should not have to piece together the most important facts from vague marketing copy.

Your content therefore has to hold up in a conversation. What would a customer ask next? Why is the offer suitable, and when is it not? What are the limits? How does it differ from the alternatives? To answer those questions clearly, you first have to understand the content yourself. If your team cannot explain an offer in a straightforward conversation, it will also struggle to structure it in a way that people and machines can use.

Trust Is Decided Before the First Click

During my solid perfume search, I did not initially place my trust in a brand. I trusted the AI that was evaluating the brands for me. That creates a new chain of trust: The person trusts the agent, the agent evaluates the available information, and the brand has to make sure that information is accurate, current, and understandable.

Trust does not come from a claim on the homepage. Customer logos, a polished design, and sentences such as “We stand for the highest quality” are not enough. Trust comes from specific statements, consistent information, and evidence that can be checked. For an AI agent, brand authority is not simply something a company declares. It is inferred from the available context. 

When your website contains unclear, contradictory, or outdated information, the agent has to make assumptions. Another brand that explains its services more clearly and supports its claims with better evidence may be easier to recommend.

SEO used to be one of those “Yes, we should probably do that” topics. It mattered, but it was often left to marketing, editorial teams, or an agency. With AEO and GEO, the conversation has reached the boardroom. Answer engine optimization and generative engine optimization are not only about ranking. They raise a more fundamental question: Does our company still appear in the answers and recommendations at all?

The terminology may change again. The underlying issue will not. Your brand needs to be understood, represented accurately, and considered trustworthy by AI systems.

The Interface Moves. The CMS Stays.

When research and advice move into an AI assistant, the website loses its role as the only interface. The interface might be a chat, a search engine, a voice assistant, or an agent inside a purchasing platform.

The content still has to come from somewhere. It has to be maintained, structured, and updated. There has to be a clear understanding of which information is valid, how different pieces of content relate to one another, and in which context they can be used. This is where the CMS remains important.

Deane Barker used a strong image for this at CMS Connect: The central content layer gets richer while individual channels become thinner or multiply. In this model, the CMS is no longer mainly the software someone uses to assemble individual pages. It becomes the trusted content source behind many different interfaces.

For that to work, it is not enough to divide content into a headline, body copy, and an image. A good content model also explains what a piece of information means, how it should be used, and what purpose it serves. A product is more than a product page. A service is more than a text field. Pricing, audiences, requirements, features, contacts, and evidence should exist as structured information—not because structure is a goal in itself, but because it makes content easier to understand, reuse, and trust.

The visible interface may move elsewhere. The CMS remains the foundation.