The new search mode on Google strangely resembles Pinterest … but in addition powerful

While this functionality very criticized by the press is not yet available in France and in other European countries, Google has just announced a major development of its AI mode.

Indeed, in a press release published on its site The Keyword, Google added Adding images to the results generated by AI. Launched last spring in the United States as a mainly textual tool, this AI mode update should make it possible to offer more complete answers, especially in visual inspiration, but also in the online shopping experience …

More immersive exploration thanks to the image

According to Google, some requests cannot be easily described with words, but more with images. As we can see in the firm’s demonstration videos, Google is focusing on a multimodal experience where the visual takes a central place.

Users can now ask a question in a conversational way and get a series of images corresponding to the desired universe, As we could see on Pinterest.

Each visual is clickable and returns towards additional sourceswith the possibility of Specify its research without starting from scratch.

Google has also planned to allow Internet users to start their search by uploading an image or taking a phototo give even more freedom to exploration.

A new way of shopping

While Openai has just announced its arrival in the world of e-commerce, the company could quickly be caught by Google. With sound Shopping Graphwhich lists more than 50 billion references Updates in real time, Google wants to transform the search for products into a conversational experience.

Rather than going through many filters on online sales sites, Google wants to simplify the experience where the user can describe what he wants to have mode, with the possibility of refining requests.

The results then display products in the form of a list, with the possibility of going to the e-commerce sites offered.

A technical feat supported by Gemini 2.5

Behind this evolution, Google relies on Gemini 2.5, its multimodal AI model, as well as on already proven technologies as Lens and the search for images.

On this occasion, Google announced that it has developed a called method « Visual Search Fan-out »capable of finely analyzing the content of an image, including its secondary details. This makes it possible to launch several requests in parallel to return more relevant results, while keeping the context sought.

With this update, Google intends to respond to an evolution of uses, where research is no longer limited to the text, but becomes A dialogue enriched by the imagethought for both inspiration and purchase …