Google this week announced the public rollout of a beta Gemini feature called Personal Intelligence. It gives Gemini access to your Gmail, Google Photos, YouTube, and Google Search data, if you grant permission, which you can do individually (for example, you can grant permission for YouTube access but not Gmail). 

The rollout will take place over the coming weeks, and the feature is exclusive to English-speaking Google AI Pro or AI Ultra plan holders in the U.S. 

What that means in practice is that when you ask Gemini for help planning a trip, it knows your flights, which airlines you have status on, what your seating preferences are, and so on, and thus can offer more relevant help. 

If you want advice on getting a car repair, it knows the make and model of your car, where you live, and your car’s history of repairs. 

And it improves upon the already pretty good search in Google Photos by enabling you to search for pictures with specific objects in them, or even dredge up specific account numbers or amounts in documents you’ve photographed. 

Google claims that Personal Intelligence does not use your data to train its models. The company also says it doesn’t make a copy of your existing stored data, but rather leaves it where it is in the current encrypted state. When the system extracts data to answer a query, that data is protected by Application Layer Transport Security. 

Google’s data advantage

You likely have a Google account, and it’s probably packed with personal data. In fact, Google probably stores more of your personal information than any other company by far. 

Consider that Google theoretically “knows”:

• Every word and number in your email for as long as you’ve used Gmail.
• All your contacts and relationships through Contacts. 
• Where you live and work, and everywhere you’ve gone since using Google Maps. 
• Everyone you’ve take pictures of with Photos. 
• The contents of all your Google Docs, as well as spreadsheets and slides. 
• Everything you’ve ever searched for in Google Search, and which results you clicked on. 

And a lot more.

>If you downloaded a personal assistant app that requested permission to access all this data, you probably wouldn’t grant it. But Google’s access to your personal data was granted long ago, and you’re used to the fact that it already has that access. 

Contrast Google’s situation with Amazon’s. The eCommerce giant acquired San Francisco-based AI wearables startup Bee in July 2025. Bee makes a $50 bracelet (which can be also worn as a clip) that listens and converts the words it hears into insights about your life, plus gives you a to-do list and reminders. (It also requires a $20-per-month subscription.)

Amazon says it hopes the Bee will deliver the kind of personal data harvesting outside the home that Alexa does inside the home. Compared to companies like, say, Apple, which has high trust among customers for the protection of user data, Amazon’s trust level is very low. 

I even have a friend who was an active Bee user, but stopped using it when Amazon bought the company. 

Google’s hardware and platform advantage

Google has another big advantage: Glasses. The company has probably spent more time and money on research and prototypes for smart glasses than any other tech firm. And it plans to release two AI glasses products this year — 12 years after being first to market with Google Glass. 

And Google rolled out its Android XR operating system last May. Android XR is Google’s glasses and goggles operating system for both its own and partners’ hardware. The company is already on the Preview 3 version of its Android XR SDK. 

Google is working with XREAL on its tethered Project Aura glasses, which are more akin in usage to Apple Vision Pro than to Meta Ray-Ban Display glasses. Project Aura is expected to result in a shipping product late this year. 

The company is also partnering with Samsung on a headset and glasses project often called Galaxy XR or Project Moohan. 

And while Google is fast-tracking products to market, its DeepMind group is doing some of the best research in the industry with Project Astra. The goal is to create a seamless, wearable, multi-modal, spatial, and personalized AI wearable that uses Gemini to harvest video, sounds, location and other context to help guide you all day. 

I’ve been predicting for something like 13 years that Apple and Google would rise to dominate the future of smart and AI glasses — just as they have done with smartphone platforms. 

Hardware design, apps and other features are important. But one of the most under-appreciated factors is access to personal data. 

Apple has super high user trust, meaning if it asks for access to user data, users will be more inclined to grant access. But Google barely even needs to ask permission. It already has nearly all the user data. Plus, it has one of the top three AI models. 

Yes, there will be many AI glasses on the market. But the most successful ones will almost certainly be those made by Google and Apple, or running on Google software. 

Other AI wearables are going nowhere

Meanwhile, so many companies looking to introduce AI wearables are lost in the weeds, nursing delusions that the public will want to wear pins that poke a hole in their clothing and hang on their shirts, or will wear a necklace that swings around and looks ridiculous. 

Fortunately, I believe CES this month represented the last surge of AI wearable delusions. The onslaught of AI pins and necklaces showcased at the event in Las Vegas included Memories.ai‘s Project LUCI multimodal pin, the Plaud NotePin S note-taker, Motorola’s AI necklace prototype, and the Nirva AI companion pendant. 

These products will never go anywhere. 

As the future of all-day, every day AI glasses comes nearer, we can now safely make a few predictions. First, that future will be centered glasses, and to a lesser extent, watches. 

Second, permission to or access to personal data will make or break the products, determining how useful and agentic they can be. 

And third, based on the above, it’s going to be Google and Apple that dominate the AI glasses market for the rest of the decade, at least.