Follow ZDNET: Add us as a preferred source on Google.
ZDNET's key takeaways
- Personal Intelligence scans your apps to improve Gemini answers.
- The feature is off by default and won't train on sensitive data.
- It's only available for paid users at first.
The Google ecosystem of apps spans everything from the cornerstone of your workflow, like Gmail, to personal apps like YouTube and Photos. Now, the company is letting Gemini connect across your entire ecosystem for more tailored assistance.
Google is calling this new capability Personal Intelligence, launched on Wednesday. Using Gemini 3, Personal Intelligence can reason across your apps to develop context that can be used to provide proactive insights. Gemini already references information from your apps, but Personal Intelligence is meant to pull more tailored responses from that data. For starters, the experience will connect Gmail, Photos, YouTube, and Search.
Personal Intelligence
In simple terms, the feature gives Gemini more of your data, which should improve its reasoning and in-practice understanding of your needs when you query it. Rather than a new interface, it's more like an upgrade running in the background. That's similar to what Apple Intelligence pitched itself as: an always-observing assistant that gathers insights from your phone activity to better answer queries.
Some examples of Personal Intelligence in action include pulling specific details from your email or photo gallery that could be helpful in answering your question or connecting an email thread to a video you watched on YouTube, according to Google.
Also: I tried Gmail's new Gemini AI features, and now I want to unsubscribe
Gemini will reference where it pulled information from, and will not use your data to personalize every answer; instead, it will only do so when it is helpful and relevant. It is also built to avoid proactive assumptions about sensitive data, such as your health, Google added.
You can also decide not to have Gemini personalize certain answers by using the “try again” button, and if unhappy with a result, share feedback, which will help refine the experience.
In particular, Josh Woodward, VP of the Gemini app, Google Labs, and the company's AI Studio, said in the blog post that while the company tried to minimize mistakes within the beta version of Personal Intelligence, there are still risks of inaccurate responses or “over-personalization,” in which the model connects two topics that are unrelated. In these instances, collecting user feedback is especially important.
Also: Gemini vs. Copilot: I compared the AI tools on 7 everyday tasks, and there's a clear winner
Also, if it makes an incorrect assumption — for example, that you are still in a relationship because of the photos in your library, or that you like golf when you're really just there to support a friend — you can correct it via chat with a simple clarifying statement.
Privacy
Of course, with an experience like this, privacy is a major concern for users. Google noted that Personal Intelligence is off by default, giving users complete control over whether they want to connect any apps, or select from some and not others.
Also: I've tested Gemini, ChatGPT, Copilot, and others – Lenovo might have all AI assistants beat
Google also said that Gemini will not train directly on your Gmail inbox or Photos library. Rather, Google clarified that it trains on limited information, such as specific prompts in Gemini and the model's responses, not details it finds in your apps, like a photo of your license plate in your camera roll.
“We don't train our systems to learn your license plate number; we train them to understand that when you ask for one, we can locate it,” the company explained.
How to access
For now, Personal Intelligence is currently only available to paid subscribers in beta, including Google AI Pro and AI Ultra users; however, Google plans to expand the feature to the free tier in the future. It is intended to reach all eligible users by the end of this week.
Once an eligible user enables it, Personal Intelligence can work across Web, Android, and iOS with any model available in the Gemini model Picker. To turn it on, open Gemini, click Settings, then Personal Intelligence, and select Connected Apps. It is only available for personal Google accounts, not for workspace business, enterprise, or education plans. Soon, it will also be coming to Search in AI Mode.
