Android is being prised open for rival voice assistants — just not here
Home Assistant has spent years running into the same wall on Android: your phone will happily listen for "Hey Google" all day, but it won't listen for anything else. In Europe, that wall is coming down — and how it came down is as interesting as the change itself.
On 16 July the European Commission adopted a decision under the Digital Markets Act requiring Google to open eleven Android capabilities to rival AI assistants on equal terms. The Open Home Foundation, consulted during the process, called it a big win for interoperability — fair enough, given the list reads like its own wish list.
What's actually being unlocked
The detail matters more than the headline. The eleven features include always-on wake word detection, ambient sensor access, screen automation, the long-press gesture that summons an assistant, structured access to apps like Gmail, Calendar and Maps, on-device AI models, and fair background-execution rules. Crucially, Google can't gate any of it behind holding the default-assistant role.
Translated into a house: you could say "Okay Nabu" and have the operating system itself hear it — efficiently, without a screen awake and without flattening the battery by lunchtime — while Gemini stays the default for everything else. That's the difference between a voice assistant that works and one you fight.
The catch, and it's a big one
- It's a European remedy. The DMA governs Google's conduct in the EU. Nothing in the decision requires the same access in Australia, and there is no Australian law in force that would compel it.
- It's slow. All eleven features must be available in Android 18 by 1 August 2027. Concurrent wake-word detection — two assistants genuinely listening at once — isn't required until Android 19, by 1 August 2028.
- Compliance is a floor. Being ordered to expose an API is not the same as it being pleasant to use. The first version will be the minimum that satisfies the Commission.
Our honest expectation is that most of this does eventually land here, because maintaining two meaningfully different Androids is expensive and Google generally doesn't. But that's a prediction, not a commitment anyone has made, and "sometime after 2027" is not a plan you can install.
What we'd do in the meantime
The practical answer hasn't changed — if anything this decision confirms it. Don't build your voice control on a phone.
- Put the wake word on hardware you control. Dedicated voice satellites — a small mic-and-speaker box on your own network — do wake-word detection locally and answer to your automations, not to a platform's permission model. They aren't waiting on a regulator.
- Keep the phone as a remote, not the brain. The Home Assistant app is genuinely good at dashboards, presence detection and notifications. Ambient listening is precisely where the platform limits bite, so don't design around it.
- Judge Google by what it ships globally, not by what it's ordered to ship. When Android 18 lands, the question to ask is whether the assistant APIs are available on an Australian handset — not whether they exist somewhere.
Why it matters outside Europe anyway
The precedent is the real story. A regulator has now treated wake-word access, sensor access and background execution as infrastructure rather than as features a platform owner may hand out at will. That framing travels, and it strengthens every open-source project that has quietly accepted second-class access to the device in your pocket because there was no alternative.
For a Melbourne household the immediate takeaway is smaller and more useful: the parts of a smart home you control outright — local hubs, local voice, local automations — are the parts that don't depend on a competition ruling in Brussels going your way. That's the same reason we specify local-first gear that keeps working when the NBN drops out.
If you want a house where voice control answers to you rather than to whoever owns the operating system, that's the way we build them.
Source: Home Assistant — A big win for Android interoperability