Google on January 14 began rolling out a feature called Personal Intelligence for its Gemini assistant that, with the user’s permission, draws on their Gmail, Google Photos, YouTube history and Search history to produce answers tailored to their own circumstances. The feature is launching as a beta in the United States for subscribers to Google’s AI Pro and AI Ultra plans.
Google describes the update as a step toward assistants that know not only about the world but about the person using them. Users connect each Google app individually and can disconnect any of them at any time, the company says.
How Personal Intelligence works
Once apps are connected, Gemini can combine information from several sources in a single answer. In an example Google gave, a user asking about replacing car tires would get a response that reflects the specific vehicle they own, drawn from past emails, documents or photos, rather than a generic explanation. The company says the assistant can pull details from text, images and video together to build the answer.
The same approach applies to recommendations. Google says suggestions for books, shows, clothing or travel can be based on a user’s actual interests, past trips and photos rather than on general trends. In one demonstration described by the company, Gemini looked at family photos and earlier trips to suggest avoiding tourist-heavy spots, proposing an overnight train journey and specific board games for the ride.
Availability and rollout
Personal Intelligence is available on the web, Android and iOS for US subscribers on the AI Pro and AI Ultra tiers. Google says it is starting with a limited group so it can learn from early use before extending the feature to more countries and, eventually, to the free tier. The company also plans to bring the capability to AI Mode in Google Search, which would make search results more personalized as well.
According to Google, the feature is off by default. Users decide which apps to connect, and the company says connected data is used to inform responses rather than to train its models in the ways some users may fear. Fortune and 9to5Google both noted that the move raises privacy questions, since it asks users to grant an AI assistant access to some of the most sensitive data Google holds about them.
Why it matters
Until now, most assistant interactions have required users to re-explain their situation in every conversation, and the answers were often generic because the model had no view of the user’s real context. By connecting to the services where that context already exists, Google is betting that Gemini can save time and give more accurate help, while asking users to trust it with more of their personal information in return.
The launch also sharpens a competitive contrast. OpenAI’s ChatGPT offers memory and connectors to third-party apps, but Google has a distribution advantage in the billions of accounts already tied to Gmail, Photos and YouTube.
