Create your demo

Articles Insights & trends

The CMS faces a new evolution: what comes next?

For twenty years we built websites around navigation. AI turns that logic around: from pages to intents, from menus to dialogue. On the step from CMS to Conversation Management System.

Matthias Snoei Founder, strategist

5 min read

For twenty years we built websites in WordPress and other Content Management Systems (CMS), around navigation: menus, pages and funnels determined how visitors found their way. AI turns that logic around.

Generation 1: the static website (code = content)

In the early days of the internet, code and content were one and the same. If you wanted to change a piece of text, you had to dive into the code. Maintenance was time-consuming and complicated.

Generation 2: the dynamic website (code ≠ content)

When the CMS arrived, code and content were separated. Databases delivered the content, templates determined the form. Marketers could publish without being developers. The content became dynamic, but the structure stayed static: users still had to navigate for themselves.

Generation 3: the generative website (code ≠ knowledge)

Websites are becoming generative: the interface shapes itself around the visitor in real time. Code and knowledge come apart. The page loses its central role, and the user's intent takes over.

The visitor of the future

The biggest change is in how visitors behave. Visitors come to a website with a task: they want to decide or get something arranged, and then move straight on. They expect a website to understand that.

A classic CMS isn't built for it. It puts information down and leaves the visitor to search for what they need. A Conversation Management System turns that logic around: the visitor says what they need and the system finds the knowledge and the next step that go with it. The structure of your content is no longer the starting point. The question is: what does this person want to achieve, right now?

That shift goes beyond the individual visitor. More and more interaction runs through agents: people, AI assistants or automated systems acting on someone else's behalf. What they have in common is that they come with a goal.

You can already see that change taking shape. Instead of clicking endlessly through menus, a user says: "I want to return this product." An AI assistant asks on behalf of its user: "Book a dentist appointment for next Tuesday."

Traditional websites aren't set up for this. They assume visitors will find their own way through a structure someone designed in advance. But agents, whether people or systems, work differently. They arrive with a specific intent, expect a direct route to an outcome and assume the site understands context.

The next generation of websites treats every visitor as an intent-driven agent. That could be a frustrated customer who wants a problem solved quickly, a professional looking for specific information, or an AI assistant carrying out a task. The form differs, the need is the same: to be understood and helped efficiently.

The transition: from broadcasting to understanding

A common misconception is that conversational means the same thing as a classic chatbot. Classic chatbots are usually set up as a separate piece of functionality bolted on top of a website. As a result they tend to have limited access to the knowledge, systems and processes inside an organization. A Conversation Management System starts from a different architecture. Rather than being an add-on, the conversation forms the central layer of the digital environment. It helps users reach their goal: to understand, arrange, decide or act.

In this third generation of websites, the user no longer has to search and navigate through pages and menus. Instead the user says what they want. The system recognizes what someone wants to achieve, retrieves the knowledge that goes with it, carries out actions where needed and applies business rules, from booking an appointment to help with a complex choice.

Concepts such as generative UI and outcome-oriented design fit this development. Where generative UI focuses on generating interfaces dynamically, the emphasis here also lies on organizing the knowledge, business rules and governance needed to help users reach their goal.

The CMS isn't going away. It stays essential for content management and publishing processes, but becomes part of a wider architecture in which knowledge, data and interaction logic come together. The center of gravity moves from pages to knowledge blocks, from menus to intents, from funnel to dialogue, from pageviews to task completion.

The central question becomes: "How do we help this visitor reach their goal?"

The 6 building blocks

A Conversation Management System for a generative website consists of six building blocks that work together.

  1. Conversation engine

    Manages how the conversation runs. It remembers the context (memory), earlier questions and answers, can hold longer conversations and makes for a smooth handover between AI and human colleagues.

  2. Intent recognition & routing

    Recognizes exactly what the user wants to achieve. It picks relevant details out of the question, such as products, dates or locations, and works out immediately what the right next step is or which process to start.

  3. Knowledge layer (RAG)

    The knowledge (content) is stored in connected form. Through Retrieval-Augmented Generation (RAG), current information about policies, products and procedures becomes directly usable for the AI. Every answer clearly shows its source.

  4. Skills & actions

    The system can also carry out actions. Through connections with the systems behind it, the AI can book appointments, open case files or place orders. That makes the website transactional too.

  5. Business rules & governance

    Compliance, authorization and security are built in as standard. This building block enforces the business rules and determines when a human specialist should step in (human in the loop).

  6. Analytics & feedback loop

    Every conversation produces data. The system shows where users get stuck, how answers come about, which knowledge is still missing and where processes could be better.

The first step?

The first step toward a generative website may be a mental shift: from broadcasting to understanding. Look into what visitors are trying to achieve, instead of only analyzing where they click.

Conversations show more than page analytics. You see where knowledge is missing, where processes stall and where you really help your visitors along.

So the question isn't whether the CMS disappears. The question is what role it takes on in a web that revolves less and less around pages and more and more around intents. If the visitor of the future no longer navigates but formulates, is our digital architecture ready for that?