Google is rolling out a new Gemini feature called Personal Intelligence that lets users personalize the assistant by connecting select Google apps with a single tap, according to an announcement from Josh Woodward, VP at Google Labs, Gemini & AI Studio. Launching as a beta in the U.S., the feature is designed to make Gemini more proactive and tailored by securely drawing on a user’s own information across connected services.
Personal Intelligence is opt-in and built around linking apps such as Gmail, Google Photos, YouTube and Search. Once enabled, Gemini can reason across multiple sources and retrieve specific details from a user’s emails, photos or other connected content to answer questions in a more contextual way. Google says users remain in control of which apps are connected and can disable the feature at any time.
In the announcement, Woodward described using the feature during a tire shop visit when he needed details about a family vehicle. Gemini not only surfaced tire size information, but also suggested tire options aligned to different driving needs while referencing prior family road trips found in Google Photos. The assistant also retrieved a license plate number from a photo and helped identify the vehicle’s trim by pulling details from Gmail, reducing the need to manually search while in line.
Google positioned the feature as especially useful for planning and recommendations, including travel ideas based on past trips and stated preferences. Woodward said Gemini recently helped plan an upcoming spring break by analyzing family interests and trip history to propose alternatives to “tourist traps,” including a suggested overnight train journey and activities for the trip.
Google emphasized privacy and transparency as core design goals. The company says Personal Intelligence is off by default, and that Gemini will attempt to reference or explain which connected information informed an answer so users can verify it. Users can also request non-personalized responses within a chat, or use temporary chats that do not apply personalization.
Google added that it does not train Gemini directly on a user’s Gmail inbox or Google Photos library. Instead, it trains on limited information such as user prompts and Gemini’s responses, with steps intended to filter or obfuscate personal data. The company also said Gemini is designed to avoid making proactive assumptions about sensitive topics like health unless a user explicitly asks.
Access is rolling out over the next week to eligible Google AI Pro and AI Ultra subscribers in the U.S., Google said. The feature will work across web, Android and iOS, and across models in Gemini’s model picker. It is expected to expand over time to more countries and to the free tier, and is also planned for AI Mode in Search. At launch, it will be available for personal Google accounts, not Workspace business, enterprise or education users. Users who don’t see an invite on Gemini’s home screen can enable it via Settings by selecting Personal Intelligence and then choosing which apps to connect.
