They Said the Website Was Dead, The Agents Disagree

A man holding a microphone presents on a stage beside a large presentation screen.

Key Takeaways

  • The website isn’t dead. AI is changing who it serves and what it needs to do.
  • Bots now make up over half of web requests, so websites must serve AI agents too.
  • AI answer engines still use your website as the authoritative source on your business.
  • Generating a site with AI is easy. Securing and maintaining it for years is not.
  • DXPs must evolve toward structured content, unified customer data, and governed AI.

The numbers are becoming difficult to ignore. Stack Overflow, once an essential destination for developers, has seen the number of new questions fall by roughly 80% year over year. Meanwhile, Cloudflare reports that automated agents and bots now account for more than half of all web requests.

For some, the conclusion seems obvious: people are no longer visiting websites because AI is giving them the answers directly. Why browse through ten pages when an assistant can summarize everything in seconds? And if AI can also generate an entire website, why would companies continue investing in sophisticated digital experience platforms? 

I understand the reasoning, but I believe it leads to the wrong conclusion. We're not witnessing the death of the website. We're witnessing a fundamental change in what a website is supposed to do, who it serves, and how organizations should manage it. 

The website is becoming both the source and the destination 

For most of the web's history, the objective was relatively straightforward. Attract visitors through search engines, bring them to your website, and guide them toward a conversion. Businesses invested heavily in SEO, design, content, and personalization because the website was where the entire customer journey happened. 

AI is changing that relationship. Increasingly, customers begin their journey through an answer engine or an assistant that researches products, compares providers, and recommends solutions before anyone visits a company's website. But there's something important missing from the argument that websites are becoming irrelevant because of this. 

Where does the AI get its information?

An answer engine still needs reliable sources, and for most organizations, the website remains the most authoritative source of information about their products, services, expertise, and policies. If anything, the quality, structure, accuracy, and accessibility of that information become more important when machines are interpreting it on behalf of potential customers. 

We're also seeing the emergence of technologies such as WebMCP, a proposed standard that allows AI agents to interact directly with website functionality rather than simply reading pages. An agent could eventually search a product catalog, request information, or complete a task through structured interactions with a website. That changes the website from something people only browse into something software can actively use. 

And we can’t forget about people. Even if automated traffic represents the majority of requests, that doesn't mean people have disappeared or that every bot represents a potential customer. When people do visit, the experience needs to offer something substantially better than an AI-generated paragraph. Personalization, interactive services, relevant recommendations, and experiences built around individual needs will matter more than ever. 

Why generating a website is different from running one

Another argument I hear increasingly often is that AI will make content management systems obsolete because anyone can generate a website through natural language instructions. As someone who started his career in software development, I find this argument interesting because it confuses the ability to create software with the responsibility of operating it. 

Generating an application may have become remarkably easy, but maintaining it over several years, keeping it secure, adapting it to changing business requirements, and ensuring it works reliably across an organization is an entirely different challenge. 

There are also economics to consider. A website generated through AI isn't free simply because the initial development takes minutes. Every modification, regeneration, integration, and troubleshooting session consumes resources, including AI tokens and human oversight. Over time, the accumulated cost can become substantial, potentially exceeding the license of a platform that already provides those capabilities in a maintained environment. 

But the financial argument is only part of the picture. A DXP provides control that’s fundamental: a governed source of truth that serves both people and machines. It gives organizations control over their content, customer data, permissions, workflows, and digital experiences without rebuilding those foundations every time requirements change. 

It also provides accumulated knowledge about how digital projects should operate. Good platforms don't only offer functionality. They help organizations adopt established practices, improve their digital maturity, and avoid repeating mistakes that software vendors and their ecosystems have already solved. 

Most importantly, the difficult part of enterprise software is understanding the requirements, integrating existing systems, managing security and compliance, maintaining reliable hosting, and ensuring that the solution continues to serve the business as it evolves. 

AI can accelerate development considerably, but it doesn't eliminate those responsibilities. 

The DXP must evolve with customers  

Defending the relevance of digital experience platforms doesn't mean defending the way they've traditionally operated. If customers increasingly interact with brands through multiple channels, devices, and AI assistants, platforms designed primarily around managing individual web pages will need to evolve. 

I believe the direction is toward connected, coordinated, and increasingly proactive digital experiences. Instead of treating every interaction as an isolated event, organizations need to understand the customer across channels and respond based on their needs, history, and context. 

That requires several capabilities working together. Content needs to be structured and reusable across websites, applications, and AI-powered interfaces. Customer data needs to contribute to a consistent profile rather than remain scattered across disconnected systems. Personalization must become more relevant, while governance ensures that automated decisions remain secure, transparent, and accountable. 

AI will increasingly participate directly in these workflows. At Kentico, we're exploring this direction through AIRA and our Agentic Marketing Suite, which help marketers with activities ranging from content strategy to customer journey optimization. KentiCopilot addresses another part of the equation by reducing the technical effort needed to create and maintain digital solutions. 

These are examples of a broader shift, not a destination in themselves. The objective is to make technology easier to operate while giving organizations greater control over increasingly complex digital experiences. 

And as implementation becomes faster, potentially reducing certain projects from weeks to days, the value of a DXP will increasingly come from the capabilities and reliability it provides over the lifetime of a digital project. 

The next visitor might not be human

I don't believe the website will disappear, just as mobile applications didn't eliminate the web, and online shopping didn't eliminate physical stores. What changes is how people interact with businesses and what they expect technology to do for them. 

For business leaders, this requires a different perspective on their digital presence. A website can no longer be evaluated only by how many visitors it attracts or how effectively its pages convert. It also needs to be accurate, structured, accessible, and useful to the AI systems increasingly acting on behalf of customers. 

That creates new responsibilities for software vendors, agencies, and the organizations they serve. We need to build digital experiences that work equally well for humans making decisions and agents helping them make those decisions. 

So rather than asking whether websites will survive AI, I think leaders should start asking a more practical question: if an AI agent can't find, understand, and use your website, how will your business reach customers who rely on them? 

The website isn't dead. But the audience it needs to serve is changing, and we need to prepare for that now.

Frequently Asked Questions

No, websites still matter, but their job is changing. AI assistants now often research, compare, and recommend products before a customer ever visits a company’s site, and bots account for more than half of web requests, according to Cloudflare. Those assistants still need reliable sources, though, and a company’s website is usually the most authoritative source on its products, services, and policies. The website now serves two audiences: AI systems that read it on a customer’s behalf, and people who still expect a better experience than an AI-generated summary.
AI assistants combine many sources, but your website is usually the most authoritative source on your own products, pricing, and policies. If that content is inaccurate, unstructured, or hard to reach, AI tools may misread your offering or rely on less accurate third-party sources instead.
You can generate a website with AI quickly, but running it for years is a much bigger job. Keeping a site secure, updating it as business needs change, connecting it to other systems, and keeping it reliable all take ongoing work. Each change, regeneration, or fix also costs AI tokens and someone’s time, and over time that can add up to more than a platform license. A digital experience platform (DXP) gives you managed control over content, customer data, permissions, and workflows, so you don’t rebuild those foundations every time something changes.
WebMCP is a proposed standard that lets AI agents use a website’s features directly instead of only reading its pages. With it, an agent could search a product catalog, request information, or complete a task through structured interactions with the site. This turns your website from something people browse into something software can use on a customer’s behalf, so being ready for agents could become a real business advantage.

Make sure AI agents can find, understand, and use your site, alongside the human visitors you already serve. Start with structured, reusable content that works on your website, apps, and AI interfaces. Bring customer data together into one consistent profile, and use governance so automated decisions stay secure, transparent, and accountable. Judge your site by more than traffic and conversions. Also ask how accurate, accessible, and useful it is to the AI systems that act for your customers.

Get ready for your upgrade

Upgrading from Kentico Xperience 13 to Xperience by Kentico gives your marketing and development teams a unified and future-proof platform with native AI that reduces complexity, and helps you create more impactful customer experiences.

Share this article

Cookie consent

We use necessary cookies to run our website and improve your experience while browsing to provide you with relevant information in your searches on our and other websites. The additional cookies are only used with your consent. With your consent, we may also transmit certain personal data to marketing platforms for targeted marketing purposes.

Configure