Google Lens now lets you search by recording video and asking out loud

Imagine you are at an aquarium watching a school of fish dart around, and you have no idea what they are called or even what keywords to type into a search bar.

Google just rolled out a massive batch of updates to Search, turning video clips and natural voice into primary search inputs.

Visual search is already booming for the company. Google noted that Lens now handles nearly 20 billion visual searches every month, with users aged 18 to 24 driving the fastest growth.

Point, record, and speak

First teased at I/O earlier this year, video understanding in Lens is now officially live for testing.

Inside the Google app on Android and iOS, you can open Lens, hold down the shutter button to record a video of moving objects, and speak your question out loud. For instance, you can point at the tank and ask, "Why are they swimming together?" Google's custom Gemini model processes the visual motion and your voice together, returning an AI Overview alongside relevant web links.

If you prefer still photos, you no longer have to type follow-up questions either. You can simply hold the shutter and ask with your voice. Both voice inputs are rolling out globally in English, though video requires enrollment in Search Labs.

Shopping with your camera is also getting smarter. If you snap a photo of a backpack at the airport, Lens connects advanced AI with Google's Shopping Graph, which houses data on over 45 billion products. Instead of just showing similar-looking bags, it identifies the exact item and pulls up reviews, retailer pricing, and store availability.

Meanwhile, Circle to Search can now identify background music from social media feeds or streaming movies without switching apps, expanding its footprint across more than 150 million Android devices.

AI reshapes the results page

Google is also changing how results are displayed once you search.

Starting this week in the U.S. on mobile, searches with open-ended answers, like recipe ideas, will feature full pages organized entirely by AI. Instead of a traditional list of links, articles, videos, and forum discussions will be clustered into thematic sections.

To address publisher backlash over AI Overviews cannibalizing web traffic, Google is rolling out an updated design globally that embeds prominent source links directly inside the generated text. In testing, Google claims this format boosted clicks to original sites. At the same time, sponsored ads are officially arriving inside AI Overviews for U.S. users.

Why it matters

Google is pushing hard to turn search into an ambient, real-time tool rather than a typing chore.

For everyday users, the friction of describing the physical world with text is disappearing. Being able to wave your camera at a broken pipe, a weird insect, or a passing outfit and just talk makes searching feel like asking a knowledgeable friend standing next to you.

For the AI landscape, this is Google leveraging its unmatched hardware distribution and data graphs. Competitors can build powerful chat interfaces, but Google has Circle to Search on 150 million phones, 20 billion monthly Lens queries, and a 45-billion-item shopping index tightly wired into local retail.

We think Google's product execution here is clever because it meets mobile users where they already are, while smartly slipping ad monetization into AI summaries. However, the long-term risk for the web ecosystem remains unresolved. Even with prominent citation links, reorganizing entire result pages with AI risks keeping users locked inside Google's walled garden, leaving publishers starving for the direct visits that keep their content alive.