Search Live: Camera-and-Voice AI Search Changes What 'Content' Means

At I/O 2026 Google introduced its biggest search box redesign in more than 25 years — one that accepts text, images, files, videos, or Chrome tabs as inputs — along with "Search Live," a conversational, multimodal way to search. The input to search is no longer just typed words, which changes what "content" an answer engine can actually read and cite.

What did Google change?

Google rebuilt the search box so it takes far more than typed queries — images, files, videos, and even open Chrome tabs can now be the input — and added Search Live for conversational, back-and-forth multimodal search. It is the biggest change to the search box in over 25 years, and it reframes search as something you do with a camera and voice as readily as a keyboard.

Why does this matter for AEO?

Because a multimodal engine can only cite what it can parse. A video with no transcript, an image with no alt text, or a chart with no labeled data is a blank spot to a model — even if it is your most valuable content. Being citable now extends beyond writing clear prose to describing and structuring everything else so an engine can extract meaning from it.

What should you do about it?

Make every non-text asset legible to a machine. Transcribe videos and audio, write real alt text, caption images, and label data clearly so the meaning is in the markup, not just the pixels. This is extractability applied beyond text — and it depends on access, making sure AI clients can actually reach and read those assets.

For the foundations, start with what multimodal search means and the extractability pillar.