Last week, I read reports of Sonos creating its own voice assistant for the company’s audio speaker products. My first reaction? “That’s just stupid,” I thought. What’s the benefit of doing that when there are already several capable digital assistants? Indeed, Sonos integrates with two of those already: Amazon Alexa and Google Assistant. Over the next 24 hours, as my brain marinated on this news, I found my stance had changed. Maybe, Sonos is on to something. Just maybe we shouldn’t be talking to our smart homes or our digital assistants. Instead, perhaps we should be talking directly to our connected products.

At this point, you’re probably the one thinking, “That’s just stupid.” I hear you. But now hear me out on this thought exercise.
Based on The Verge’s report, Sonos’s voice assistant is expected to deliver two things. More user privacy and a better experience are the likely results:
“In keeping with Sonos’ interest in privacy, the feature will not record user audio commands or relay them to the cloud for processing. “Hey Sonos” will be the wake word for Sonos Voice Control, and the company’s internal tests show it to be quicker than competing assistant services at core music tasks.”
If that’s indeed what a Sonos voice assistant brings to the table, that’s a win in my book. You’d get locally processed voice commands and more immediate feedback from the speakers.
Hmm…. local interaction and immediate feedback. That reminds me of something from the pre-smart home days. Devices that you interact with to have them instantly respond to your wishes.

Think about it. Before the days of wireless light switches and voice commands, if you wanted to turn a light on, what did you do? (Or what do you still do when your smart home system is on the fritz?) You interact directly with a physical switch and you see immediate feedback. Oh and nobody outside of your home knows you’re in the same room as that switch or that you’ve flipped it one way or the other.
This pre-2010 scenario, in combination with the Sonos news, makes me wonder: Should we be talking to our connected devices a little more directly? Should each device or device type have its own digital assistant?
It sounds crazy, given the smart home of today. We associate our smart home itself with Alexa, Google, or Siri and dole out tasks for the many connected devices we have. And it works, of course.
It can also be confusing, however. How many times have you, a family member or a guest fouled up a voice command simply because you or they didn’t know the exact device name? My wife can’t figure out the lights in my office, for example, because I have two lamps with three bulbs each. I named them in a way that seemed logical to me: Bulbs on the left lamp, have “left” in their name, for example. And then each bulb has an additional modifier based on its lamp position: top, middle, or bottom. I get it, but my wife doesn’t. Nor would anyone staying at our home.

But if my wife could speak directly to the lamp — well, indirectly through a smart speaker that piped commands to the lights related in the office — she could have more control over the lights. Saying “Hey lights…” gets the attention of the right devices the first time. And from there, a smarter, device-focused assistant and algorithm could take over.
The same model applies to other device types too. Why does anyone need to say, “Hey Siri, lock the front door?” Maybe the smart lock is called “front entrance” and you simply forgot. I think I’d rather say, “Hey door, lock up.” when in the room near the front door. At night, let me just say “Hey TV, tune to Netflix.” instead of calling in a third-party arbitrator.
Each of these device-specific assistants could be better tuned to meet our needs. How? Thanks to Matter, each device knows what it’s capable of and what its state is. Those capabilities could change as device makers build new efficiencies or new features. Sure, developers can integrate them into a digital assistant with APIs, but that’s because they have to in today’s world. Standing on the shoulders of the digital assistant giants before them, capable companies could deliver faster, better experiences with their connected devices.
I’m not suggesting that Matter is the complete solution here. It’s simply a part of the overall vision, providing device information as standardized glue to make a device-centric relationship become reality. There still needs to be some central processing and message routing in the smart home. Our smart speakers and digital assistants currently fill that role, as do dedicated smart home hubs. A generic hub to handle device-specific interaction would be required. And for the smaller device makers that may not have the resources or the chops to build device-centric functionality, there’s still room for Amazon, Google, and Siri.

This solution could better deliver on the promise of a smarter smart home. My lights may notice that I dim them whenever I turn on the TV at night. Why can’t the “smarter” lights I’m envisioning say, “Hey, I noticed that we get dimmer when you turn on around 7 or 8 pm, TV. How about you shoot us a message the next time this happens and we’ll just dim when you do your thing?”
Now you have self-created routines and a better relationship with your connected devices. Today’s approach is comprised of centralized algorithms, at best, to understand what’s going on in the many smart home “transactions”. Zeroing in on specific device-to-device transactions or better yet, human interactions between devices might surface more value.
Yes, this is what the smart speakers and their cloud services are supposed to deliver, so it’s a similar but slightly different paradigm. And frankly, I don’t think the centralized services have completely delivered on the smart home promise. A more targeted set of services, based on device type, could change that.

The approach is similar to what Stacey recently wrote about in her post “The smart home doesn’t need a robot — it should be a robot”, by removing one of the obstacles she notes: “A home has so many functions and is a highly individual environment that a general-purpose humanoid robot doesn’t make sense.” For all intents and purposes, today’s digital assistants are general-purpose, not equipped to handle the intricacies of many unique connected device types.
As someone whose first marriage ended in divorce and offers too much sarcasm to family members in his second marriage (I’m working on it!), perhaps I’m not the best person to give relationship advice.
Still, I think I’d rather have a direct relationship with the connected devices in my home as much as, if not more so, than with my smart home. A voice assistant for each device type just might be the way to court that engagement. Sonos just might be right after all.
Seems like this would make it more complicated, complex and expensive. Its better to have a smart hub that knows what you want to call something and have your voice assistant work with it. Asking Alexa to turn on the silver lamp, then she asks home assistant to turn on what is called Silver Lamp. Then home assistant turns on the zlgbee bulb in the $9 silver IKEA lamp.
Imagine the demands to have the silver lamp being able to track and understand when someone in the room with multiple other devices are talking to it. Not to mention the privacy concerns for the homekit crowd.
No thank you.
I agree with @LawrenceK , this means more things to remember, not fewer. Because then each person has to remember the name of each device.
Alexa makes this easy: you can walk into any room and say “Alexa, turn on the lights” or “Alexa, lights on” and whatever lights you set up to turn on in that room will turn on. Very simple.
At our house we are three housemates with lots of friends coming and going as well as health aides and other service people. It’s really easy to tell people to just say Alexa, lights on. But if they have to know that the table lamp is called silver lamp, not gray lamp or table lamp or sideboard lamp or whatever, and they have to know that for every device, they would give up very quickly.
I don’t have a problem with expensive systems like Sonos or Ecobee thermostats building in a voice assistant. Choice is good. That might be the first smart device that anybody got. Or the only one. If they want voice control and it’s built-in, go for it.
But for a whole house solution, having each device have a mandatory voice assistant just doesn’t make sense to me. Because what I call that little lamp on the red table might not be what you call it. Or the next visitor.
Optional, sure. Again, choice is good, and people can set things up the way that makes sense to them.
My guess is, from a business standpoint, that Sonos is aiming this at people who buy Sonos before they automate the rest of their house and they want voice control with it. Then it makes sense to me. But we’ll see how it works out for them.
I agree with JD Roberts and with LawrenceK, and for all the same reasons. In a home full of multiple devices in many rooms, the Alexa apps make it easy to address them individually or in groups as one prefers. Why make things more complicated by forcing each household member to memorize the name of each individual item in the house? And why add to the cost of manufacture of each electrical device by having to build a voice assistant into it and requiring each device to learn to recognize each resident’s voice when Alexa, Google or Apple do that once for all? And who would have the patience to go around the house teaching each and every light bulb etc. to learn their voice?