This article was AI-generated as part of an experimental historical-content project. The date reflects the period being analyzed rather than the date the article was originally written.
For two years, a new Gemini model followed a familiar route. It reached the Gemini app and developers first, and Search adopted it later, with less fanfare. That changed yesterday.
Google introduced Gemini 3, and Sundar Pichai wrote: “This is the first time we are shipping Gemini in Search on day one.” According to Elizabeth Reid’s post, Google AI Pro and Ultra subscribers in the U.S. can use Gemini 3 Pro by selecting “Thinking” in AI Mode’s model menu. Google says it will bring Gemini 3 in AI Mode to everyone in the U.S. soon, and in the coming weeks will start routing the most challenging questions in AI Mode and AI Overviews to the new model, beginning with Pro and Ultra subscribers.
The overlooked part
Most coverage will compare benchmarks. The more consequential change for companies is the coupling. A model launch is now also a Search launch. When Google ships a frontier model, the answers people get about a company inside Google’s own search experience can change the same day, at least for some users, and more broadly as routing rolls out.
Two details in Reid’s post deserve attention.
The first is fan-out. Query fan-out breaks one question into many searches, and Google says it can now “perform even more searches” and find content “it may have previously missed.” Retrieval for questions about a company will reach further into the web. That can surface a better source the old system skipped. It can also surface the obscure page you hoped nobody would read.
The second is generative UI. Gemini 3 in AI Mode can build layouts on the fly, with images, tables, grids and even interactive tools coded for the query. Google’s example is a loan calculator generated inside the answer. For brands, a comparison of your product against competitors may arrive as a table the model designed, with columns it chose. A table looks more authoritative than a paragraph, and errors inside it are harder to spot.
Google also makes a factual-accuracy claim: Gemini 3 Pro scores 72.1 percent on SimpleQA Verified, which it describes as state of the art. That is a benchmark result and worth reading as one. It says little about how the model handles a particular company, especially where sources disagree.
Why the timing matters
Search is where the largest number of people meet a company. Pichai’s note says AI Overviews now reach 2 billion users a month. When a model upgrade lands there on day one, the old rhythm of reputation work, where you tested a new chatbot weeks before it touched mainstream search, no longer holds.
In late 2023 I wrote that Gemini might change who summarizes you first, even if it would not replace Search overnight. Two years later, the model and the search engine are released together.
What to do this week
Rerun your core company and executive questions in AI Mode. If anyone on the team has a Pro or Ultra subscription, run them again with Thinking selected and compare the two. Save both with dates. Look for tables and comparisons, and check every cell. Then watch for the routing change Google described, because when it arrives, AI Overviews for harder questions about your company may shift too.
Model launches used to be technology news. For anyone managing how a company appears in Google, they are now search news.