
Google Assistant is being permanently removed from Android starting September 4, 2026. Not a soft sunset, not a legacy mode — a forced removal from Android phones, tablets, Wear OS watches, headphones, and phone-projected Android Auto. Once the transition hits a device, there is no switching back. Gemini takes over. If you have integrations built on top of Assistant’s APIs or voice behaviors, you are out of time to treat this casually.
What Is Actually Being Removed
Google’s rollout begins September 4 and is expected to take several weeks to reach all users. Affected platforms: Android phones and tablets, Wear OS watches, paired headphones, and Android Auto when projected from a phone. The exception is cars running Google Built-In hardware — those retain Assistant for now on a separate, unannounced timeline. Google Home speakers, smart displays, and Google TV are on their own tracks as well.
The removal is permanent. Google confirmed there is no opt-out and no rollback path once Gemini replaces Assistant on a given device.
This Is Not a Rebrand
The most important thing developers need to understand: Gemini is not Google Assistant with a different name. They are architecturally different systems. Google Assistant was deterministic — a rules-based command-and-response engine with strict logic trees. The query “set a timer for five minutes” triggered the same code path every time. High consistency. Predictable behavior.
Gemini is an LLM. It is probabilistic. The same voice query can produce different responses across sessions. Renascence put it plainly: “task-completion reliability will register as a service regression, regardless of how much more capable Gemini is in open-ended conversation.” That is the tension every developer building voice-first or voice-dependent apps now has to manage.
What Breaks for Developers
Several integrations will break or degrade without action:
- Interpreter mode — Gemini does not have it yet. If your app relied on Assistant’s real-time translation capability, that feature is gone.
- Third-party music, podcast, news, and radio integrations — Gemini lacks some of the service integrations Assistant supported. Check your specific category.
- Android Auto voice experiences — Gemini is replacing Assistant in phone-projected Android Auto, but the integration was incomplete as of mid-2026. Some users will hit gaps before Gemini reaches parity.
- App Actions (Built-in Intents) — App Actions declared in shortcuts.xml continue to work, but Gemini’s probabilistic handling means fulfillment behavior may differ from what Assistant produced. Test before assuming compatibility.
What You Do Not Need to Touch
Not everything requires migration work. Works with Google Home (Cloud-to-Cloud) integrations get the transition for free — Google’s developer blog stated that “the core conversational benefits require no additional work from you” for existing integrations. Matter-based smart home integrations are unaffected at the protocol level. Google Built-In car integrations continue as normal for now.
If your app relies only on App Actions via shortcuts.xml and BIIs, you likely do not need to change code — but you absolutely need to run behavioral tests against Gemini before September 4.
Your Migration Checklist This Week
- Audit your integrations — Inventory every place your app surfaces voice functionality: App Actions, Android Auto, smart home, any Conversational Actions (already deprecated since June 2023)
- Run behavioral tests against Gemini — App Actions may continue working, but verify Gemini’s fulfillment behavior matches your users’ expectations
- Migrate music, podcast, and news integrations — Check the Gemini API documentation and Firebase AI Logic for replacement paths
- For smart home developers — Google explicitly advised to “robustly test your existing Works with Google Home integrations immediately.” Do not wait.
- For Android Auto — Test your car mode voice experience under Gemini and identify any functional gaps
- Migration APIs to know — Gemini API at ai.google.dev, Firebase AI Logic for Firebase apps, Agent Development Kit (ADK) for building full agentic voice experiences on Android
The Bottom Line
Google is not asking for permission. The September 4 date is firm, the transition is irreversible per device, and the architectural shift from deterministic to probabilistic voice handling is real. The developers who treated this as a distant concern are the ones getting surprised this week. Audit your integrations, run your tests, and know which migration APIs apply to your stack — before your users find the gaps before you do.













