Google Integrates Personal Intelligence into AI Search

A week after introducing a new personalization tool for its Gemini chatbot, Google LLC today brought the technology to its flagship search engine. The feature, which is called Personal Intelligence, is initially available to U.S. users with Google AI Pro and Ultra accounts. Google AI is a subscription offering that gives consumers to the company’s most capable artificial intelligence models. It also provides certain other benefits including additional cloud storage space. The version of Personal Intelligence that debuted last week enables Google’s Gemini assistant to customize its responses based on users’ personal data. Consumers can give the chatbot access to that data by connecting it to Gmail and Google Photos. Google says that it won’t train its models on the files users share with Gemini. The version of Personal Intelligence that started rolling out to Google Search today works in much the same way. It’s part of AI Mode, a feature that was added to the search engine last year. It generates more detailed answers to user prompts than the standard interface. Under the hood, AI Mode uses a technique called query-fan out to generate responses. It breaks down each prompt into sub-topics and researches them separately with a large number of AI-generated queries. Going forward, the technology will be capable of pulling information from not only the public web but also users’ Gmail inboxes and Google Photos libraries. Google says that the upgrade will enable AI Mode to personalize multiple aspects of its prompt responses. “If you need a new coat for your upcoming trip, AI Mode could automatically take into account the brands you prefer, as well as your flight confirmation in Gmail to identify the destination and timing (Chicago in March),” Robby Stein, the vice president of product for Google Search, wrote in a blog post. Personal Intelligence and AI Mode’s fan-out query mechanism are both powered by Gemini 3, Google’s newest series of large language models. AI Mode routes the most complex user queries to the top-end Gemini 3 Pro algorithm. Simpler requests are processed by the other models in the series, which uses less infrastructure. AI Mode personalizes its output based not only on users’ preferences and files but also the data types a given prompt response contains. If an AI-generated answer contains structured numerical information, the feature can display a spreadsheet table. It’s also capable of generating more complex graphics such animated star system simulations. AI Mode’s data-specific interface customization is powered by a version of Gemini 3 Pro that has been enhanced with third-party tool integrations and error correction features. There are several ways Google could enhance AI Mode’s implementation of Personal Intelligence in the future. The company can add integrations with more services besides Gmail and Google Photos, including third-party applications. Additionally, it could make the technology available to enterprise users. Developers, for example, might benefit from a version of Personal Intelligence that tailors coding-related search results to the project they’re currently working on. Image: Google