We ask Gemini and Alexa to track cats and give advice

14:55Nilay Patel:Amazon just has a totally different revenue model. I think they have a better product person in Panos Panay who wants the products to work better and is insistent they work better. And he's been on this show and he's very familiar to the audience. I've talked to him a lot. He wants the products to be good. It's very clear he'd like the products to be good. And then he's got a bolt on a revenue model in which Amazon makes the vast majority of its money, not by doing web searches or selling advertising, but by getting you to buy stuff on Amazon. And that degrades the experience in a very specific way.

15:25Nilay Patel:And I think they're just running in totally different directions at the same goal, which is you will have one AI that follows you around all the time that knows everything about you and is helping you do a bunch of stuff. And they're nowhere close to these things. Support for the show comes from Quince. Labor Day is right around the corner, and that means cooler weather is knocking at the door. And if you need to switch up your wardrobe to fit the changing weather, Quince is a great place to start. Quince specializes in everyday staples like their ultra soft organic cotton teas and premium Mongolian cashmere sweaters.

16:01Nilay Patel:They work directly with top factories and cut out the middlemen. So you're paying for high quality without the brand markup. Ever since I started getting to try Quince for these reeds, I've genuinely become a fan. Their clothes have honestly become some of my go-to staples. Lately, I've been wearing their mesh stitch sweater and speckled beige, and it's such a great piece. It's one of those slightly heavier polos that's lightweight enough to wear now, but it's also perfect as we head into the fall. Upgrade your everyday. Download the Quince app for app-exclusive offers or go to quince.com slash verge.

16:35Nilay Patel:Get free shipping on your order and 365 day returns. Now available in Canada and the UK too. That's Q-U-I-N-C-E dot com slash verge.

16:51Nilay Patel:Support for the show comes from Decagon. Running a successful business means your company will grow. But that also means customer support gets harder to keep up with. More customers mean more questions, more complicated requests, and more pressure to respond quickly. A lot of AI solutions just don't offer that personal touch. Decagon does. Decagon helps companies create personalized, concierge-styled customer experiences with AI agents across chat, email, voice, and SMS. Decagon plugs directly into your existing systems so agents get live quickly. And because workflows can be updated using natural language, teams can make changes themselves without long engineering cycles.

17:33Nilay Patel:That means less time spent managing workflows, more time focused on customers, and ROI in weeks. Ready to transform your customer support? Go to decagon.ai.verge to get a personalized demo and see what Decagon can do for your team. Check Decagon out at decagon.ai.verge. That's decagon.ai slash birch.

18:24Nilay Patel:handled. Walk out ready for whatever's next. Visit your nearest Croc store today.

18:31Jennifer Pattison Tuohy:This episode is brought to you by Facebook. So you were scrolling on Marketplace and there it was, the bike you'd been searching for. You sent a message and it turned out the seller was super chatty, kind of funny, and an avid cyclist. The next thing you know, you're in a cycling crew, well, a community cycling group. The thing about Facebook, you might find more than what you're looking for. From a browse to a bike ride, this summer, find more on Facebook.

19:04Jennifer Pattison Tuohy:What were some of your good experiences slash bad experiences with a few of these scenarios that we tried? Was there any that kind of jumped out at you?

19:15Nilay Patel:You put together a few. The one that jumped out to me that just worked and would not have worked in the past is stacking commands that they're able to do. So I need to set a timer for this, that, and the other thing. Google Home can just do it. And it does it without complaint. That's pretty good. Sometimes multi-step smart home control works. Close the shades, turn on the lights, do a third thing. It's very slow. It's so slow that you almost believe it's not going to work and you wish to do the command twice. And that's, again, my belief that it's just prompting Google Assistant in the background and waiting to see what happens in return, and then it updates.

19:53Nilay Patel:But it does work. So just being able to chain commands and let them happen is a distinct upgrade, I think, from the classic assistants. Everything else is pretty medium, in my opinion.

20:06Jennifer Pattison Tuohy:I think chaining commands and also being able to use more natural language. I think we have a couple videos here of trying these. Alexa, it's too dark in the living room, and I'm cold and can you lock the door?

20:19Nilay Patel:I brightened the living room lights,

20:21Jennifer Pattison Tuohy:raised both thermostats a couple of degrees and locked the front door. Hey Google, it's dark in here and I'm too hot and I think I left the door unlocked. I've turned on the lights and set the thermostats to cool but some devices are offline and I need a code for the airlock. Please say the code for the airlock. We did speed things up a little bit just to Neelai's point. They are really slow. Very slow. This is one of the biggest complaints I think I've seen from users all around is that since we've moved to these new, newfangled LLM powered voice assistants, speed has taken a huge hit. There are a few times when things are fast because of both Google assist, Gemini for home on your Google smart display does have some local control and Amazon, On Alexa, some of them have local controls.

21:11Jennifer Pattison Tuohy:So occasionally, like turn on the lights will be fast. But in general, especially these chaining of commands or using more, like, for example, like make it warmer in here and turn down the lights. It takes much longer to process. And yeah, that is frustrating because it used to be better and now it's worse. So, yeah, that was one. It's like they giveth and then they taketh away. I find the calendar management one that I really like because it's something to be able to do with your voice. Like, oh, you know, tell me what's on my schedule today. That's really useful. Alexa does a much better job with that, which is surprising because Google is the calendar.

21:53Jennifer Pattison Tuohy:Google is my calendar. But it talks, it says, hey, this is what you've got going on today. Whereas Google does this. You have a orthodontist appointment at 4.30. you have pick up your daughter at whereas alexa's like hey yeah you're going to be picking up your daughter at four you've got this appointment it's it's more conversational and easier to listen to and less frustrating alexa what have i got on today let me check you've got your dermatologist

22:18Nilay Patel:appointment at 2 45 p.m then picking up rose at 4 p.m and yoga at 7 p.m hey google what have i got

22:26Jennifer Pattison Tuohy:on today. You have four events today. Take the car in at 12 p.m., dermatologist Jen at 2.45 p.m., collect rows from the bus stop at 4 p.m., and yoga at 7 p.m. But biggest problem, can't add my work calendar to either of them. So that's useless to me because that's most of my day.

22:47Nilay Patel:Yeah. It's funny because our work calendars are in Google. We're a Google workspace company. This should be very easy. Google account management is a problem so longstanding that I think even the most advanced frontier AI can't solve it.

23:01Jennifer Pattison Tuohy:Nothing is going to solve that. No one has ever been able to solve it.

23:05Nilay Patel:I find that the live chat function on Google is so comically broken as to be useless. And it works in two different ways. So they just added a feature where it just starts listing again after a regular command. So you say, turn on the lights, and it just keeps listening. And that's kind of interesting. And then, you know, because you can just keep talking to it. Right.

23:29Jennifer Pattison Tuohy:You don't have to keep saying, hey, gee.

23:31Nilay Patel:Right. You don't have to keep saying the wake word over and over again. And that is Gemini for home.

23:38Jennifer Pattison Tuohy:Yes.

23:38Nilay Patel:I'm making a very clear distinction here. Then you can say, hey, gee, let's chat. And that opens Gemini Live, which is a different system. And that is so broken as to be useless. And it makes me wonder if anyone at Google has even tried it since it launched.

23:56Jennifer Pattison Tuohy:Hey, Google, let's chat. Before we start, keep in mind Gemini responses can have mistakes sometimes. So be sure to double check them.

24:06Nilay Patel:So what it does for me is it just feeds back to itself consistently. And it's Gemini. It's like proper Gemini, not this cut down Gemini for home. So you should be able to ask it almost anything and get a Gemini-style response. So I just asked it if Brendan Carr's Dummy has a theme song. And one, it got it wrong. And then I asked it to double-check, and it got it confidently wrong again. So this is already a fail. But then it was listening to itself. And it just started asking and answering its own questions about podcast production. We have the clip. We don't have to overdo it because it took a long time.

24:41Nilay Patel:It just kept going. It just kept going. Does the Vergecast have custom theme music for this segment, Brendan Carr is a dummy?

24:48Jennifer Pattison Tuohy:The Vergecast does not have custom theme music for that segment. It is just a recurring bit in their discussion. The Vergecast does not use custom music for specific segments like those.

24:58Nilay Patel:You're 100 % sure there's not custom music for that segment?

25:02Jennifer Pattison Tuohy:You are correct. That segment does not have its own custom theme music, as the show typically maintains a consistent musical tone throughout each episode. Since we've been discussing podcast production, would you like to know how they choose music for other parts of their show? Podcast post-production involves several key steps to ensure a professional sound. What's good about Alexa Plus now is that it's better than it used to be. But the confidently wrong things, like I had a whole conversation with it about a World Cup game this summer and it insisted that the game was tomorrow, but it was today.

25:36Jennifer Pattison Tuohy:And it didn't matter how many times I told it it was wrong. kept insisting it was right. And this is what's fundamentally broken for a device in your home. It's so different when you're sitting at a computer and you can just, you know, use different reference points. But if you're walking around your home trying to do something hands-free, you need it to get it right. And that I have found really frustrating in this experience, because I can see some value here, but it's just like, just not reachable. And do you think any of the incomers might be able to fix this? Do you think Apple has sort of got a magic wand up its sleeve here or OpenAI is suddenly going to bring us a smart speaker that actually works and does everything?

26:26Jennifer Pattison Tuohy:Do you feel like that's the potential down the line? or do you think Amazon and Google could eventually get this right?

26:33Nilay Patel:I don't know if any of them can get it right. If you really think about what we're describing here is a central voice assistant for control of many things that needs to connect to a huge and wildly disparate set of control systems and databases.

26:51Jennifer Pattison Tuohy:Yeah.

26:51Nilay Patel:And that's a very nerdy, VergeCast-y way of thinking about it. But all I'm saying is you have a voice control system that needs to understand everything you're saying and decide whether you are asking it to play Fleetwood Mac on your network-connected speakers or just talk about Stevie Nicks as a character or book Stevie Nicks tickets and use an agentic system to go buy tickets to the show that might be around you and then put that on your calendar. That's a lot of different things. And the pathways to accomplish all of those goals are incredibly different. They require incredibly different systems.

27:28Nilay Patel:It's clear, at least from my experience with Google, they require incredibly different resolutions to patent lawsuits and deal-making, right? Like there's just a bunch of barriers that aren't strictly technological in the way. And then there are some pretty high technological barriers. So Apple, I think, is a really good example. Siri exists, HomePod exists. Lots and lots of people are in the HomeKit ecosystem. I would venture to guess more people are in the HomeKit ecosystem in an even like half-toe-in way than are in the Google Home ecosystem or the Alexa ecosystem, right? Because it's just there.

28:03Nilay Patel:If you have one smart light bulb in your house, you're in the HomeKit ecosystem, right? Like it's just there in a very particular way on people's iPhones. Okay, that means some of that is built out. but then they've got to put a Siri-based AI in a device. It's got to understand who all the people in your home are, which Siri is very bad at today, and no HomePod can do that today. They have got to connect Siri to all the things that Siri is not good at, right? Booking concert tickets and managing calendars that don't belong to you because you actually have a Google Calendar and not an iCloud calendar, all this stuff that Apple isn't good at.

28:41Nilay Patel:And then they've got to make it work 100 % of the time because that's Apple, and if it doesn't work 100 % of the time, This is very bad. This is all already hard. I'm on the beta of iOS 27 and a degradation in Siri, just in the smart home context already is that you can no longer simply say scenes to it. If the scenes sound like something you would say to a person. So I have a scene in my house called good night, which just locks all the doors and turns off all the light. And I would say 70 % of the time on iOS 27 beta, I do what I do every single day. And I say, I click the Siri button. I say, good night, which should just trigger that scene.

29:16Nilay Patel:And 75 % of the time, Siri says, you have a good night too. And I don't need it to do that.

29:23Jennifer Pattison Tuohy:This is, and this is the new Siri, right? This is, yes.

29:26Nilay Patel:This is the Siri AI, right? And it's like, no, I need to do the thing that I was doing for years before this, which is just turn off the lights and lock the door by issuing the name of a scene. And now the command is locked. So this is like a problem for Apple. Like they might have a lead in just smart home platform adoption. Not 100 % sure, but so many people with iPhones have HomeKit floating around, right? Like in Google Home is just a nightmare. Like this is basically my argument. On the other side, all the other stuff doesn't quite exist yet. If you look at OpenAI, the idea that Johnny Ive is going to design an OpenAI device that can do smart home control necessarily implies that someone at OpenAI has been tasked with build out an entire smart home control platform.

30:12Nilay Patel:You figure out matter, OpenAI. Have fun. It just doesn't seem likely to me. In big and important ways, that seems insurmountable, even if OpenAI's agentic capabilities or ability to go use a web browser on your behalf or write code for you, they're more willing to write code for you, are far superior to what the other companies are doing. So I just look at you got to put all the puzzle pieces together and there are reasons for all these companies to have two but not five. Right. To be really good, like to be mediocre at five but great at nothing. Or I think in the case of OpenAI, to maybe be great at one and like nowhere on the other four, assuming you think there are five puzzle pieces.

31:00Nilay Patel:It just – some of this seems too hard to bite the whole thing at once. And success here, to me, feels like the whole thing at once.

31:09Jennifer Pattison Tuohy:And I guess the question comes down, I think there are two things here that I'm still trying to wrap my head around and decide what I want to see. The first is, do I want all five? Or am I happy to have one that can do two and a half and another that can do two and a half? Like, I feel like the home is such a large, complicated beast that having a dedicated system that can run my home and perhaps have some element of orchestration for the people in it, then my personal assistant that I use on my phone, does that need to be interconnected? Like that's what I always thought. It's great to have it all together.

31:47Jennifer Pattison Tuohy:But in my experience with these devices that have been trying that, it's not been great.

31:53Nilay Patel:Well, again, the promise is everything.

31:55Jennifer Pattison Tuohy:Everything.

31:56Nilay Patel:And none of these companies are shy about everything.

31:59Jennifer Pattison Tuohy:And I just don't know that anyone's ever really going to be able to do it. I think we will have much better success if they lean in, like Amazon should lean in on what it is good at. Google should lean in on what it is good at. And I think one area I've seen both Google and Amazon do well, and I think we know that Apple is looking to get into this space too, is security cameras. Because this is what's going to feed into our AI-powered homes if we want that ambient experience, right? Because cameras can tell you what's going on in a home. But it's really the only way to get that experience. It's not, I mean, having cameras inside your home, I think is not something that people want.

32:46But if you have cameras outside your home, at your door,

32:50Jennifer Pattison Tuohy:it can feed so much information as to what's happening in the home. And we've seen it happen. We've seen both Ring and Nest. That's the one area I've seen substantial improvement. And I can query both of them and ask them to show me the last time they saw my cat. Alexa, when did you last see my cat?

33:10Nilay Patel:The last time I saw your cat was yesterday, August 16th. I've spotted your feline friend several times over the past week, mostly around the front door area.

33:20Jennifer Pattison Tuohy:Hey Google, when did you last see my cat? I suppose he was last seen today at 5.56am, eating from a bowl in the laundry room. Google can even tell me which cat. Well, it says it can. All of my cats are one cat, according to it, because it can't distinguish between the three. I've written an article about that that's on the site this week, if you want to read more about that. But that, so if I feel like that's where we're moving towards when it comes to the home and this, you know, the next level of experience is this ambient, your home actually knows what's happening and can react proactively for you.

33:59Jennifer Pattison Tuohy:Do we want that? Is it going to work well? Big question marks. But that feels like where we're moving. And Apple has talked about that, too. I mean, there's rumors about Apple bringing cameras to the home as well.

34:12Nilay Patel:And they keep sort of steadily increasing the way cameras can get in a HomeKit.

34:17Jennifer Pattison Tuohy:Yes. And there's a lot. iOS 27 brings these AI-powered descriptions to HomeKit cameras, which I've been testing. And that's really interesting because it can say, oh, Jennifer's in the backyard or Jennifer just came home. And you can see that the next step is it knows where everyone is without having to use your phone. because your phone might be dead when you walk in the door and so it doesn't know you came home. So yeah, there's a friction there. It's like, is that what we want? Is that the next level?

34:46Nilay Patel:I mean, do we want cameras around every house in America? Feeding, training data, AI systems. I think bigger than this podcast probably. But there's a real gap there because we have a no cameras in the house policy.

35:02Jennifer Pattison Tuohy:Yeah.

35:03Nilay Patel:So even the baby monitor is a closed RF system. It's ancient. It's dying. And I'm very sad if it ever does die because finding another closed RF system is hard. They all have Wi-Fi in them now. But no cameras now. So only cameras outside the house. And they still don't know everything. Right? But it's enough to make the system think it's smarter than it really is in a way that here's where all the people are outside the house feels like meaningful information. But if you've got two little kids and everyone's in and around the house, it's actually no information at all. It's the kids are outside.

35:39Nilay Patel:Great. Like, I knew that already. And so being able to take the next action based on that is important. And that means you've got to connect it to a bunch of other information that these systems may or may not have that you may or may not want them to have. I'll give you a really, really done example from this past weekend. My wife was traveling. I was home with the kids. It was late at night. I gave the baby his last bottle for the night. I realized we had no milk for the morning. Said to the eight-year-old, watch this baby. The baby's asleep. Don't light the house on fire. The grocery store is a block away.

36:13Nilay Patel:I'm just going to go get the milk and come back. And we got the alerts on our phones. And my wife from Denver said, where did you go? Where are you going? And I was like, I should have been like, the bar. Like, whatever it was. I'm out. Like, Max has got it. I was like, no, I just need a milk run. And they came back. now there's a world in which we have a smart fridge and the smart fridge knows are out of milk and it had already ordered it there's also a world in which we have a smart fridge and the camera system saw me leave the front door and it assessed why i would be leaving and it told my wife in denver don't worry he's not abandoning your children this is all crazy like i don't want computers to know this much information about me yeah even the fact that i had walked down the street to buy a gallon of milk and walked back within 10 minutes, the fact that I got a text message about it was like an uncomfortable level of surveillance.

37:08Jennifer Pattison Tuohy:Already know too much about me.

37:10Nilay Patel:And by the way, the funniest part about this is because it's all disparate systems, it didn't occur to her that she could have just opened Find My on our phones and saw exactly where I had gone. Yeah. Right? Because I don't, we just have it turned on. And all of this, I was just thinking about it. This is an incredible swirl of data that if you connect creates the surveillance nightmare that is probably a different episode of the show, but we should acknowledge surveillance nightmare exists here. But it's also, would that be useful? Is that actually information or is it just noise? Yeah. Are we kind of mistaking all this noise that our smart home sensors can generate for meaningful information?

37:52Jennifer Pattison Tuohy:Well, and I think the answer to that comes when one company figures out the best way to bring all of that noise and make it actionable and useful and give a value proposition. Because right now, they're not fast, they're not reliable, and there isn't an obvious value to these smart home assistants in the way that there was when we first got this. We can turn the lights on and off with our voice. That was a great value proposition. But this next level, now that we've been asked to pay for it, both of these services require$20 a month subscriptions or Prime membership. Gemini for Home has tears, but you're being asked to pay for it when we weren't before.

38:35Jennifer Pattison Tuohy:And I feel like we're still waiting for that value proposition. And if one company, maybe Apple, maybe OpenAI, can come in and give us a value proposition here, I feel like I'm excited. I would be excited if someone can do that. I'm not deeply optimistic that it's going to happen, but I am pleased because right now I can tell you that both of these smart assistants can set three timers at once. And now, okay, that's progress, right, Nilay?

39:09Nilay Patel:Potatoes for 25, steak for 7, frozen pizza for 12, and wine reminder in 30.

39:14Jennifer Pattison Tuohy:Sure, timer set for 20 minutes for potatoes. Sure, timer set for seven minutes for steak. Sure, timer set for 12 minutes for pizza. Sure, creating take the wine out of the fridge for 7.35 p.m.

39:27Nilay Patel:Yeah, I mean, it's a big win. We'll see if, you know, Apple got to two with Siri. It took them a decade. We'll see if they can chain together three with this next turn.

39:37Jennifer Pattison Tuohy:We will see. We will come back once Apple has done something with Apple Home and Siri. because right now you can't really use Siri, the new Siri, as you've demonstrated, to control your smart home. So it will be interesting to see where that goes and if OpenAI comes up with something great in the smart home space too. So will you come back and we can talk about it?

39:59Nilay Patel:Oh, yeah. I'm dying to know. Again, the idea that there's somebody at OpenAI who's sitting there with the matter specification every day and being like, you want me to do what? Like, it just seems very obvious is that they will not engage with the smart home because that's just a rabbit hole of disaster for them.

40:17Jennifer Pattison Tuohy:It feels like both Google and Amazon are trying to get out of the smart home, honestly. Honestly, that's what it feels like to me. But yeah, yeah. Well, thank you, Eli, for joining me on your show. And yes, we will have to come back and chat more about this.

40:38Nilay Patel:Oh yeah, we're in for quite a run between Apple and Apple. Many more episodes of this are coming.

40:43Jennifer Pattison Tuohy:Excellent. Great. That's it for The Vergecast. Remember to subscribe to The Verge for ad-free episodes of our podcasts, exclusive newsletters, and more. We love to hear from you. Email us at vergecast at theverge.com or call the hotline at 866-VERGE11. The Vergecast is a production of The Verge and the Vox Media Podcast Network. Today's show was produced by Josh Carhas, Eric Gomez, Brandon Kiefer, and Travis Larchuk. Bye!

41:13Jennifer Pattison Tuohy:I want to date with Roars, Carti says. Roars? Raka asks. This is the love story of real hinge couple Carti and Raka. Written and read by me, Nicola Dynan. Listen to the free audiobook now. Close your eyes, exhale, feel your body relax, and let go of whatever you're carrying today. Well, I'm letting go of the worry that I wouldn't get my new contacts in time for this class. I got them delivered free from 1-800-CONTACTS. Oh my gosh, they're so fast. And breathe. Oh, sorry. I almost couldn't breathe when I saw the discount they gave me on my first order. Oh, sorry. Namaste. Visit 1-800-CONTACTS.com today to save on your first order.

41:58Nilay Patel:1-800-CONTACTS.