An APK teardown of the most recent Android Auto beta suggests a useful addition that might be on the way: Gemini-powered shortcuts that can be pinned to the car’s launcher for single-tap functionality.
The new feature seems intended to declutter the sort of common-but-not-final actions once offered by Google Assistant, possibly supplanting gangly Actions with more open-ended, richer commands handled by Google’s generative AI.
Although the feature isn’t available, strings hidden in Android Auto version 15.6.154404-release suggest that folks who can get the app update will be those who see it.
Drivers will be able to add and name a Gemini action, attach the dictated command to it, and then drag and drop it on an app drawer with tiles that users are already familiar with. Consider it like the speed of pressing a physical preset button for AI-driven requests.
What the Android Auto Teardown Reveals About Gemini
Language in the beta mentions adding a Gemini action and entering a command for Gemini, which is described as incorporating “one-tap” navigation and also refers to calling a contact. That’s similar to today’s Assistant Actions workflow on Android Auto—Customize launcher > Add shortcut > Assistant action—but with Gemini as the engine. However, the UI does look like it allows testing the command and selecting an icon label, although none of this has been enabled in the beta build.
The language suggests two shortcut categories:
- Direct-dial tiles for calls
- Gemini tiles for AI-triggered actions
How Gemini Shortcuts Could Function in the Car
What makes Gemini valuable here isn’t voice recognition as such; it’s orchestration. A single shortcut could initiate a multi-step task that traverses across apps and context. Examples could be:
- “Send my ETA to Alex on WhatsApp and initiate navigation home.”
- “Play my Focus playlist on Spotify at a low volume.”
- “Find an open EV charger near my destination and add a stop.”
With Gemini’s action chain, they can be taken care of with just one tap — no distractions.
Google has already indicated that Gemini will make its way to the car experience with options for message summaries and suggestions based on context. Shortcut tiles would inch that service from reactive voice prompts into the realm of proactive, repeatable workflows — think habitual schedules like school drop-offs or daily gym runs.
Why It Matters for Safety and Usability on the Road
Distraction is an ongoing hazard on the road. U.S. safety regulators have been reporting thousands of annual deaths that they attribute to distraction, and research from groups like the Virginia Tech Transportation Institute has shown that even momentary glances away from the road can increase crash risk by wide margins. Cutting down on-screen taps and cognitive load is important in real cars, not just spec sheets.
One-tap Gemini presets support best practices: minimizing interactions, reducing glance times, and providing predictable outcomes. So look for Google to institute driving-adapted limitations like short responses, low visual content, and adherence to narrowly defined allowances in the same way that Android Auto already restricts activities behind the wheel.
What Happens to Assistant Actions as Gemini Arrives
The teardown lingo here indicates Gemini will be where Assistant Actions currently live, implying that there may be a replacement or the possibility of migration between platforms.
In practice, that could look like current custom commands getting a Gemini upgrade under the hood (retaining users’ launcher layouts while now allowing more natural language and multi-step logic). It also fits into Google’s larger shift to unify its consumer-facing assistants under the Gemini brand, instead of splitting them across phones, tablets, and web.
One open question is how much runs on-device vs in the cloud. Android already supports on-device AI in Gemini Nano for some experiences, but Android Auto usually gets its smarts from the connected phone and network. You can most likely expect some sort of hybrid, with sensitive data handled under Android Auto’s already-established permissions and privacy protections.
When You May See It on Android Auto’s Launcher
As with all APK teardowns, there is no guarantee that these changes will actually see the light of day. However, the specificity of the strings suggests a feature that’s currently still in progress. It would arrive with other experiments we’ve seen through the rearview mirror, like wacky home screen widgets, multi-media cards, and the soon arrival of Call Screen and Call Notes—features all designed to reduce friction, but not create a bunch more for things that are outside of its vertical.
For drivers, the promise is simple: less to do and think about when doing the things you do every day, but smarter execution as work gets more complex. If Gemini shortcuts do ship as suggested, Android Auto’s launcher might not be so much a grid of apps, but a canvas of personalized AI-backed presets.