Google’s AI Will Group Links in Search Results

Google has introduced an experimental search tool called Web Guide. Its goal is to change the usual look of search results using artificial intelligence, which will group links by topic and generate structured answers.

The function is controlled by a modified Gemini model, which reorganizes classic search results, breaking down the query into semantic parts and selecting related topics. Thus, the user receives not only links, but also SI analysis, categories, and clarifying hints.

How it works

The new function is already available to Search Labs program participants and works within the Web tab — exactly where there are no AI Overviews yet. The approach is similar to how the AI Search mode works: the system simultaneously performs several refined searches to cover more information.

Example: If a user asks “how to care for a mango tree”, Web Guide returns:

• two relevant pages,
• a short summary of the topic,
• categories like “Care in Different Climates” or “Common Problems with Mango Trees”.

Sources the system relies on: official websites, YouTube, Quora, Reddit. (Recall that in 2023, Google signed a $60 million per year contract with Reddit for the use of content to train AI.)

Web Guide partially returns to the old Google style, with text links at the top, without intrusive videos and “vague” AI reviews. However, not everything is perfect – in some blocks, the AI adds obvious comments, such as: “The best method depends on the age of the tree, climate, and variety. Check out several guides.”

The tool can also handle complex and abstract queries. For example: “My family lives in different time zones. What tools will help us stay close?” – Web Guide will try to break the question down into components and give a detailed answer for each part.

If the user doesn’t like the new approach, they can always switch back to classic search. But Google has already hinted that in the future, Web Guide will appear on the main “All” tab. You can try Web Guide after activating it in Search Labs.