From pages to answers: Why website search needs to change
Most site search tools require visitors to think like the people who wrote the content. They type a keyword, you match it against your text, and if their word isn't your word, they get nothing useful. Synonyms break it, different phrasing breaks it, and increasingly people don't want a list of links at all. They want answers.
The problem is that most websites were built for a different era. We invested years into pages, articles, product information, documentation, and knowledge bases. But when someone wants information, we still ask them to dig through that content by hand. Meanwhile, expectations have moved on. People ask ChatGPT a question in plain language and get a direct response, and they now expect the same from websites, customer portals, documentation, and knowledge bases.
It's becoming clear that people want something better than a traditional search box. What that looks like is still an open question, though, because an AI model by itself isn't enough.
AI alone creates a different problem
Ask a large language model a question and it will usually produce a fluent, confident answer. The catch is that it doesn't know your content. It doesn't know your documentation, your product information, your articles, or your internal knowledge. Worse, it can confidently invent details that were never published in the first place.
For a documentation site, a product catalog, or a knowledge base, a confident wrong answer is often worse than no answer at all. It erodes the trust your content was supposed to build. So the challenge isn't generating answers, because modern models do that easily. The challenge is generating answers that are grounded in your real content, and that a reader can verify.
The missing piece is structured content
Most organizations already have the information they need. The problem is that their content was designed to render pages, not to answer questions.
Structured content changes that. Pages, articles, snippets, and documentation aren't only website building blocks, they are knowledge assets. Once content is structured, it can be reused well beyond rendering a webpage. It can power search. It can power AI assistants. It can power applications that don't exist yet. The same content your editors create today can become the knowledge layer behind entirely new experiences, without anyone rewriting it.
This realization has shaped much of our thinking at Sulu over the last years. We have always argued that structured content creates long-term value beyond websites. AI has simply made that value visible in a much more immediate way.
The question we started asking ourselves was simple: if structured content already contains an organization's knowledge, how can we make that knowledge accessible in entirely new ways?
This is the part we find genuinely exciting at Sulu. We've always argued that structuring content is worth the effort because it creates maintainable, scalable digital platforms. The rise of AI has turned that long-held principle into an immediate, practical advantage.
Organizations that invested in structured content are discovering that they already possess something extremely valuable: a reusable knowledge layer. Instead of rebuilding content for every new interface, they can use the same content foundation to power websites, search experiences, AI assistants, and entirely new applications.
From websites to knowledge platforms
This is the shift we're seeing across the industry. Structured content shouldn't only power websites. It should become a reusable knowledge layer for search, AI assistants, internal knowledge systems, and future applications.
The organizations that will benefit most from AI won't necessarily be those with access to the latest models. They will be the ones with well-structured knowledge that AI systems can actually use.
Instead of forcing visitors to navigate through pages, organizations can let people ask a question and receive an answer drawn directly from their content. The content stays the source of truth. Only the interface changes, from a page you browse to an answer you receive.
This shift from pages to knowledge is exactly the challenge we've been working on at Sulu. Our first step in that direction is Intelligent Search.
Intelligent Search is our first step
This is exactly what Intelligent Search is designed to do. It indexes your Sulu content and lets visitors ask questions in natural language. Instead of returning a list of results, it generates an answer grounded in your content and backed by citations to the original pages.
Under the hood, Intelligent Search first retrieves relevant information from your own content and then uses that information to generate an answer. The model supplies the language; your content supplies the facts.
In practice that means content stays under editorial control, answers stay connected to their source material, information stays current as content changes, and visitors get answers instead of link lists.
Most importantly, you've already done the hard work. You structured your content. Intelligent Search simply puts that structure to work.
If you want to understand the mechanics, embeddings, retrieval pipelines, and how a question becomes a cited answer, the next post in this series, How Sulu's Intelligent Search Works, walks through them.
What's next
Intelligent Search is currently available through a private onboarding program, where we're working with early adopters to learn from real production use cases and shape what comes next.
And there's a lot that comes next. Search is only the first step. Our larger goal is to turn structured content into an AI-ready knowledge layer across the Sulu ecosystem — one that can power assistants, developer tools, and applications we haven't built yet. Intelligent Search is the first practical expression of that vision.
If you'd like to see how it works with your own content, schedule an onboarding session. As always, questions and feedback are welcome on GitHub and in our Slack community.
