AR | Augmented Reality · Technology News
What you can build for Google's first Android XR glasses
Google’s first Android XR glasses go on sale on 10 December with a 12MP camera, open-ear speakers and Gemini but, according to Android Police, no built-in display. You can reach them with an existing Android app, whose notifications are read aloud with no glasses-specific code, provided the app uses Android’s standard notification APIs, or with a native Android build that follows Google’s camera and microphone rules. Unity, the engine we build in, is not on Google’s list of tools for them.
What Google is selling, and who can use it
Pre-orders opened on 12 November, announced by Shahram Izadi, Google’s VP and GM of XR, in Google’s launch post. The collections are made with Warby Parker, Gentle Monster and Samsung, and the Warby Parker frames come with “a wide range of prescription, width, and lens options”, which turns prescription glasses with a camera and Gemini into something you can pre-order in the US. Google’s own posts describe audio, the camera and Gemini rather than a screen; 9to5Google calls them audio glasses.
| Detail | Source | |
|---|---|---|
| Price | Warby Parker from $395; Gentle Monster from $449 | |
| Weight | Around 41 g (9to5Google gives 40-44 g) | Google, 9to5Google |
| Battery | Up to nine hours; case recharges the frames up to seven times | |
| Camera | 12MP ultra-wide; video up to 90 seconds | Google, 9to5Google hands-on |
| Chip and memory | Snapdragon AR1 Gen 1, 2GB RAM, 64GB storage | 9to5Google |
| Controls | Touchpad taps and swipes, or voice with a “Hey Gemini” hotword | 9to5Google |
| Phones | Pixel 7 and later, on the September 2026 Android 17 update; Galaxy S, S FE, Z Fold, Flip and Fold Ultra on One UI 9.0; iPhone 13 Pro and later | |
| Accounts | US and South Korean Google accounts, 18+ only |
Two details matter more than the price. 9to5Google’s hands-on says the glasses are “offloading a lot of processing to your phone”, which matches how Google asks developers to build for them. And Google’s footnote says the glasses are “only available for use with Google accounts from the United States and South Korea”. Google’s launch post of 12 November gives no UK or European date, so a UK team cannot run a pilot on consumer units yet.
Compatible or differentiated: the two ways onto the glasses
Google’s quality guidelines for augmented experiences split glasses apps into two tiers. A compatible app needs no glasses-specific code, as long as it uses Android’s standard notification APIs: “A compatible mobile app provides an augmented experience on glasses with zero additional developer effort.” On audio glasses, its notifications are read aloud. A differentiated app is purpose-built with Google’s Jetpack XR SDK, a set of native Android libraries.
| Compatible | Differentiated | |
|---|---|---|
| What it is | Your existing Android app | An experience built for glasses |
| Code needed | “No glasses-specific code is required” | Jetpack Compose Glimmer, Jetpack Projected, ARCore for Jetpack XR |
| What the wearer gets | Notifications read aloud | Voice launch, glasses camera, location-aware features |
| Where it runs | On the phone | A projected activity inside your phone app |
What matters most is the last row: glasses apps run on the phone. Google’s first-activity guide says that, unlike headsets, glasses “use a dedicated activity that runs within your phone’s existing app”. An activity is a single screen or task inside an Android app; here it is projected to the glasses rather than shown on the phone. You extend the phone app you already have (Google’s samples are in Kotlin, Android’s main language) and share its business logic. A launcher category makes the activity discoverable by Gemini voice commands using your app’s name. The libraries are still at alpha versions (Jetpack Projected is at 1.0.0-alpha14), so expect the interfaces to change.
That changes the size of the job. Checking the compatible tier is a review of the notifications your app already sends: whether each one makes sense spoken aloud, and whether it would interrupt someone at the wrong moment. A projected activity that uses the glasses camera is new software, with a voice entry point, a camera permission request, the countdown Google asks for, care over battery drain, and testing on glasses that UK accounts can’t yet use. We would treat the first as one task inside a wider piece of work and the second as a project in its own right, with its own discovery, a prototype and a re-test each time an alpha interface changes.
Google’s own example of a differentiated feature is modest: a video conferencing app that lets the user switch its stream to the glasses camera for a first-person view. ARCore for Jetpack XR adds device tracking and the Geospatial API, which Google says enable “walking navigation”.
The rules that decide most ideas
Before anyone designs a glasses feature, they should read Google’s prohibited list, which applies to every glasses app. Apps must not use the camera for “real-time, one-to-many live streaming, public broadcasting, or casting to public feeds and open channels”. They must not run facial identification “to surface bystander Personally Identifiable Information (PII)”. They must not keep “persistent, background audio recording or transcription” when the wearer is not on a task they started, and must not change, dim or disable the LEDs that show the camera or microphone is in use. The tap gesture must not become “a continuous controller for interactive games”.
What is allowed: video calls, and using the camera “for ambient camera sensing or scene understanding”. If the camera is started from the phone, the wearer should hear a countdown first. Google also asks for experiences built around user-initiated actions that minimise distraction, especially when people are moving.
On our reading, a one-to-one call with a remote colleague sits on the allowed side; broadcasting a site walk to an open channel does not. As AndroidPure puts it, “These are Google Play policies that bind app developers, not a hardware lock.” Wearers sign up too: Google’s privacy post says that before first use they have to accept rules against filming people who have not given consent or recording in sensitive locations.
Where Unity fits, and where it doesn’t
Google’s device-type page lists Unity, OpenXR and WebXR as tools for XR headsets and wired XR glasses. For audio and display glasses it lists only the Jetpack XR SDK. Google’s Unity guide says Unity’s Android XR support is built on OpenXR, the cross-vendor XR standard, and is marked for headsets and wired glasses only.
So the glasses activity itself is native Android work. That does not rule Unity out of the phone app around it. Our engineering view is that a phone app with embedded Unity content can also carry a native Kotlin projected activity, because Google describes the glasses side as one more activity in the app you already have. It has a cost: an embedded engine makes the app larger and slower to start, and one project then has to keep a Unity build and an alpha Jetpack library in step. Google’s guides don’t show the two together, so we would prove the combination in a prototype first, and only where the phone side really needs real-time 3D.
Unity does the visual work beside the glasses: phone and tablet AR with ARKit and ARCore, and headset apps for Android XR and Meta Quest. We also build the companion app, accounts and backend that a glasses pilot needs.
iPhone owners are not shut out as wearers: Google lists iPhone 13 Pro and later as supported phones, so they can pair the glasses and use Gemini. The developer guides we read cover Android apps only, so for now we would plan a pilot that depends on your own app around Android phones.
What audio-only means for location AR
Cornhill Christmas, the free outdoor AR app we built with Urban XR for Ipswich Borough Council, depends entirely on a screen. Sammy the reindeer appears on the Town Hall steps and Santa’s sleigh flies over the rooftops of The Cornhill, placed against the real buildings on visitors’ own phones.
None of that is possible on glasses without a display. What the Geospatial API could add is a spoken layer: directions to the next character, or a story told as you stand in front of the Town Hall, with the visual moment staying on the phone. Building it would start with a site check that the Geospatial API can place a wearer accurately on The Cornhill, then an audio script for each stop. And no visitor in Ipswich could hear it until Google allows UK accounts.
How we would scope a glasses pilot now
This is the order we would work in during discovery:
- Check that your Android app uses standard notifications that make sense when read aloud, since that reaches the glasses for free.
- Pick one task the wearer starts and finishes without using their hands, such as a warehouse picker asking for the next bin location: it begins with a voice request, and the answer fits in a single spoken sentence.
- Test every camera and microphone use against Google’s prohibited list before design starts. We would also run a data protection review before any pilot that puts a camera on staff or the public, including how workers are monitored.
- Build the notification layer and the shared business logic in the phone app now, so the glasses activity stays thin while the libraries are in alpha, and add that activity when Google gives a UK date. Budget for a re-test on each new Jetpack Projected release.
If you are considering a glasses pilot, get in touch: on a first call we can go through your app’s notifications with you and check your camera and microphone ideas against Google’s prohibited list.