Google is testing a more personal version of Gemini, one that can draw connections across parts of a user's Google activity instead of waiting for the user to point it toward a single app.
The new beta feature, called Personal Intelligence, starts with Gmail, Photos, Search, and YouTube history. Google says the goal is to let Gemini understand more context around a request, so its answers can be shaped by information already sitting across a user's Google ecosystem.
What Personal Intelligence changes
Gemini could already retrieve information from Google apps. The new feature goes further by reasoning across those sources. In Google's framing, that means the assistant can connect one clue from Gmail with another from Photos, Search, or YouTube history to produce a more tailored response.
The practical difference is that users may not have to specify where Gemini should look. Google says Gemini can understand context without being told the source, then use Personal Intelligence when it determines that the added context would be helpful.
Josh Woodward, VP, Gemini app, Google Labs, and AI Studio, described the feature this way in a blog post: "Personal Intelligence has two core strengths: reasoning across complex sources and retrieving specific details from, say, an email or photo to answer your question,"
He added that it can work across text, photos and video to create answers that are more specific to the user. The feature is designed less like a simple search tool and more like an assistant that can combine scattered details into a single response.
Why the examples matter
Google's examples show the kind of everyday use it has in mind. Woodward said that while standing in line at a tire shop, he could not remember his car's tire size. Many AI chatbots can help identify tire size for a car, but Google says Gemini can personalize the answer by using related information from the user's own account.
In that example, Gemini suggested all-weather tires after identifying family road trip photos in Google Photos. Woodward also said he forgot his license plate number, and Gemini was able to retrieve it from a picture in Photos.
The same logic applies to recommendations. Woodward wrote that Gemini had been useful for books, shows, clothes and travel. For spring break planning, he said Gemini analyzed family interests and past trips in Gmail and Photos, avoided tourist traps, and suggested an overnight train journey as well as specific board games for the trip.
Google also listed example prompts that show the intended range:
- "Help me plan my weekend in [city i.e. New York] based on things I like to do,"
- "Recommend some documentaries based on what I’ve been curious about,"
- "Based on my delivery and grocery receipts in Gmail, Search history, and YouTube watch history, recommend 5 YouTube channels that match my cooking style or meal prep vibe."
These examples point to a shift in how Gemini may be used. Instead of asking for generic advice, users can ask for answers based on their own habits, receipts, photos, searches, videos, and past plans, as long as those connected apps are enabled.
Controls, limits, and sensitive topics
Personal Intelligence is off by default. Google says users can choose if and when they want to connect their Google apps to Gemini. That opt-in design matters because the feature can involve deeply personal material, including photos, emails, Search history, and YouTube history.
Google also says Gemini will not use Personal Intelligence constantly. If a user connects apps, the assistant will use the feature only when it decides that doing so would help answer the request.
The company says it has guardrails for sensitive topics. Gemini will avoid making proactive assumptions about sensitive data such as health. At the same time, Google says Gemini will discuss that data if the user asks it to.
Google also says Gemini does not train directly on a user's Gmail inbox or Google Photos library. According to the company, the model trains on specific prompts in Gemini and the model's responses. In the examples Google described, the road trip photos, the license plate image, and the Gmail messages were referenced to generate a response, but were not directly used to train the model.
Who gets it first
Personal Intelligence is rolling out to Google AI Pro and AI Ultra subscribers in the U.S. Google plans to expand the beta to more countries and to Gemini's free tier.
For now, the feature is best understood as a test of how far a consumer AI assistant can go when it is allowed to connect personal context across multiple services. The benefit is more relevant answers. The tradeoff is that users must decide whether that relevance is worth giving Gemini access to more of their Google activity.
That choice is central to the product. Personal Intelligence is not simply about making Gemini smarter in the abstract. It is about making Gemini more aware of the person asking the question, using the apps and history that already shape what that person does online.