After nearly 10 years, Google’s smart home ecosystem is starting to look more than a little shabby. Not only has it been five years since Google last released a new smart speaker, but Google Assistant, the voice assistant that powers all of your voice computing, feels less effective than ever. That’s not just conjecture, either. In July, Google was looking at a potential class action lawsuit because of what I would describe as a complete and utter meltdown of the Google Home app, preventing people from turning their lights on and off and accessing their security cameras.

You may, if you are anything like me, have been tempted to write a blog during this smart home calamity, claiming “Google’s Smart Home Ecosystem Is Crumbling” or “Google Home Is So Bad a Lawsuit Could Be On Its Way,” and it would have been justified in doing so. However, what you wouldn’t know, if you wrote those words at that time, is that an alleged fix was already on its way. I’m talking, of course, about Gemini for Home, Google’s next-gen smart home assistant, which is now rolling out in early access.

Gemini for Home, in case you missed the news in September, is a smart home voice assistant that uses Google’s large language model (LLM), Gemini. It’s being billed as a more advanced version of the Google Assistant that currently lives on your current Google hardware and sometimes (God willing) successfully turns your lights off and on about 65% of the time. In other words, it’s the dawning of a new AI-fueled era for voice assistants, and thanks to the chatbot’s ability to understand natural language, it should (theoretically) be a lot less frustrating.

It’s also, if your name starts with a “G” and you run the most popular search engine in the world, time to put your money where your mouth is. As a matter of record-keeping, I am going to compile a few ways that Google says Gemini for Home is better than the Google Assistant of yore. From the mouth of Big G itself, here is how Gemini for Home has been described:

• “An AI upgrade”
• “Foundational intelligence that transforms your relationship with your home”
• “… you can have a real back-and-forth conversation without having to constantly repeat yourself”
• “Gemini can better help you across with the main ways you use your assistant today: media, household coordination and smart home control”
• “Gemini moves from being a simple notetaker to a proactive partner by understanding and interpreting the intent behind your requests for calendars, lists, timers and reminders.”
• Gemini is the “next era of Google Home”

As you may have noticed from these grandiose statements, Gemini for Home is supposed to be a big deal, and the reason I bother to highlight that is that I have my doubts about whether Google will actually be able to deliver on all of those promises. As I noted previously, harnessing the power of chatbots might be tougher than Google is making it seem—just ask Apple and its botched Siri rollout.

I’m also highlighting all of these lofty promises for an even more important reason, which is that Google is now asking for money if you want to step into the future of voice computing. While a “basic AI upgrade” for all of its existing smart speakers is free, there are more premium plans offering additional features like automation assistance (the AI helps you set automations) and more advanced camera notifications. A standard plan costs $10 a month, while an advanced plan costs $20. If you’re going to be putting your hard-earned money down on this stuff, you should probably know exactly what’s being promised.

If you’re ready to step into the future of Google-made smart homes, you can join the early access program right now via the Google Home app by tapping on your profile picture, navigating to settings, and selecting “Early Access” at the bottom of the page. Unfortunately, you’ll have to wait until 2026 to do that with hardware, though, since Google’s next Home speaker isn’t coming until the spring. Either way, Google’s made a lot of promises here, and it looks like we’ll finally get to see if it can deliver the promise of a transformative voice assistant 10 years later.