The green dress. Jennifer Lopez wore it at the Grammys in 2000. The internet didn’t break back then, exactly, but people definitely looked. Hard. By July 2001 that search demand pushed Google to launch an actual image tool. It has been 25 years. The tool changed everything, yet rarely makes the news cycle. To mark the anniversary, I talked with Lou Wang. He co-founded Google Lens. We discussed two new things arriving in your browser soon. Plus how visual search survived its own infancy.
From text to camera inputs
The early days were boringly simple. You typed words. Google showed you pictures. That was the deal for over a decade. Then came reverse search. Upload a picture or paste a URL. See where it came from. In 2017 Google made the big pivot toward visual search proper with Lens. First it lived in Google Photos and Assistant. Then it moved to the main bar.
The game changed when inputs stopped being just text. Multisearch let you mix a photo and a question. Circle to Search let you circle a part of your screen and ask about it. You didn’t switch apps. You just used the camera.
What is the new Google Images home?
If you hate hunting, you will like this one. Google is launching a browsable homepage for Images. It acts like a dynamic gallery. Real-time. Tailored to what you actually care about. But here is the utility part: tabbed collections. Think Pinterest boards or TikTok saves, but native to Search.
You can build a place to drop similar visuals. Say you are planning a trip. You want to visit Malaysia. Create a “Malaysia” collection. It shows up as a tab above your gallery. You save the street food. You save the temple roofs. It stays there. No hunting through browser history.
This lands on desktop first. In the US. English only. Coming soon.
Generating room visuals with AI Overviews
The other feature is wilder. Direct image generation inside AI Overviews. Not just text descriptions of images. Actual generated pictures. The engine here is called Nano Banana. Yes. That is the name of the model. You use text prompts.
Suppose you are redecorating. You can’t pick between colors. Yellow-gold vs. sage green walls in your master bedroom. You type that into the Overview box. Ask for a visual comparison. The AI builds the room. It paints the walls. You see which looks better without buying sample chips.
This arrives in regions that already support image creation in AI mode. Also English only, for now. Wang says user demand forces tech to stretch past its limits. They want intuitive answers. Mix voice. Text. Visuals. However the human wants to ask it.
“Why complicate it when the camera already sees the answer?” Wang asked me.
He isn’t wrong. We just started typing. Now we show.
