Hello and welcome to The Vergecast, the flagship podcast of CarPlay entitlements, which are a thing I know about against my will. I'm your friend David Pierce. Nilay is here. Hello, sir.
I know so much about CarPlay entitlements. I am so excited to talk more. I am forcing more Home View talk into the top of this episode because CarPlay is about to become more important in your life. And I think that is so funny.
That's the worst. By the way, can I just shout out Jen 2 is amazing. It was so much fun to have her on the live stream afterwards. And we saw comments that were like, 'You should let Jen talk more and she's British and polite and you just run over her.' I just like talking to Jen a lot.
Yeah, that's it. That's what I've got. Like it is fun to talk to our reporters who are deep nerds on stuff and approximately one person in this world wants to talk to me about Z-Wave and it's Jen. That's it. That's...
I just want you to know I just get excited to have another nerd who cares about my nerd stuff here. Jen does not care about CarPlay. The only other person who cares about CarPlay as much as me is Joanna Stern, whose Home View videos were like, 'CarPlay.' And then she like excitedly texted me about it.
So, I'm working on it. We're going to bring it all together. And then we'll make a show that like four people enjoy, but I will have the most...
And all four of them are the hosts of the show. It's a show for no one except the hosts.
It's going to be great. I'm excited about it. We have a lot of stuff to get to today. We have a bunch of AI news. We're going to talk about this thing where people are starting to slink back to X, which I find fascinating. We've got some SpaceX stuff going on. We've got a lot of news to get to, but I do want to start with a little bit of Home View stuff at the top here. We did the live stream on Wednesday about Apple's new smart home entrant, play here. But one thing we talked a little bit about and were a little bit confused by and got some feedback about was the app situation on the Home View. And I would say both that we've had some time to learn about it and read about it and talk to people about it, but also that I feel incredibly vindicated because it is really confusing and essentially makes no sense. So, Nilay, do you just want to as best you can sort of lay out the app situation on the Home View as we understand it now?
There is no app situation on the Home View.
Well, it's not exactly... it's mostly true. It's so true. The app situation on the Home View is approximately the same app situation if you went on to Amazon or Alibaba and bought a $299 CarPlay tablet for your car.
Yeah, that's basically right. Yeah, they sell those. You can just get one. Oh, yeah. They're in the TikTok shop all the time. Yeah, you can just get one. You just like mount it to your car. They sell some that are so big you just mount them over your vents.
I've always been taken with these.
Yeah. One day we'll do an entire episode on just people who own older model Ford F-150s ripping out the entire dash and putting in CarPlay tablets. There's a whole ecosystem there that I'm passing by.
That's basically what Apple did here. They made the Home View. They built some first-party software to do some things. In particular, make FaceTime calls and the Home app.
Yeah. It's not totally dissimilar from what you get kind of out of the box with the iPhone, right? It's like you get alarms, you get Apple Calendar, you get the Home app, you get Messages, you get Podcasts. It's sort of the bog standard set of Apple apps are on this thing, right? But there's no App Store. And then there's a bunch of Apple apps that aren't there, right? Like if you are an Apple Music Classical person, which has its own app because classical music is organized in complicated different ways, that is not available to you. Apple News, a thing you might wish to scroll on your displays while the chaos of your family makes French toast around you. Not available here. All of these things are going to be CarPlay apps. And if you will recall on the day of the Home View launch, it seemed like what Apple was going to do was take CarPlay and then like rework it for the Home View. And instead, it seems like they're just going to force everybody as a CarPlay app into supporting the Home View.
Yeah. There's a thing in the developer terms that essentially says if you make a CarPlay app, it will work on the Home View. The end. There's like you barely even have the ability to turn it off. Like if it is in CarPlay, it will be on the Home View. Does this approach make any sense to you? Because I confess this does not make any sense to me.
Uh, no. First of all, Apple's entire deal is apps. Yeah. And they're already asking a lot of developers. The iPhone Duo is going to require a lot of developers to support yet another screen size. There's some amount of rumor in the rumor mill about a touchscreen Mac. Developers are going to have to think about that in some new way.
I mean, this is a new hardware category for Apple. It is a new way of thinking about how you might interact with Apple devices in your home. And of all of the things you would ask developers to do, hey, think about like a true cool fun multi-user app for our shared screen in the kitchen. Seems like top of the list, right? This is a new category and instead Apple has chained it to the iPhone in a way that means, I mean they've sold a lot of iPhones but they've chained it into the iPhone in a way that means you can only ever sell some fraction of the number of iPhones that exist and so the app model is that and the user model is that and it's a lot of iPhones, don't get me wrong, I think most companies would be happy to sell a significant portion of a fraction of iPhones.
And I can't imagine somebody who isn't already in the iPhone world is even going to be really looking at this thing. But maybe that's kind of the point, right? Like this thing is an accessory of an iPhone. It's an Apple Watch in a very real way in that there is no reason whatsoever to buy it separately from owning an iPhone in a way that like you can own an iPad and an Android phone and live a life that makes sense. This will make no sense if you are not an iPhone family. And if you recall, and I know you recall because you wrote the Wired cover story on the first Apple Watch, which Jobs got mad at you, right?
Right? The Apple Watch started as like a remote control for your iPhone apps. Like that was a lot of the app model for the Apple Watch is, yep, this is just a little screen, but the app is actually running on your phone. And over time, as the processor got more efficient and more powerful, they brought the apps to the watch. This is why I was basically screaming, 'Where is the computer?' on the live stream this week.
The watch became a computer and then it started running the applications and now the watch is a vastly more independent computer even though it's still connected to your iPhone in a lot of ways. The Home View does not appear to be a computer, right? It is the model of that first generation Apple Watch where it's a remote for your phone is right there. And then it's on top of that, it's very funny they picked CarPlay as the app model for those little remote control apps. CarPlay. The way to think about CarPlay is that it is just a second monitor for your iPhone, right? It is screen projection and you plug it in or you wirelessly connect, your iPhone detects a second monitor that's a CarPlay app and it shares a screen with custom interfaces for the apps that are running on the phone. This is fine in a car. It's also not like it isn't fine in a car. And this is the thing like CarPlay, I am the pro CarPlay person on this podcast by default. I don't think CarPlay is very good, right? Like this experience is not good. Like the simplest example for me is Spotify. You don't get anything remotely approaching the full Spotify experience in CarPlay, which makes some sense when you're driving and you should be tapping as little as possible. But if I'm going to have to do the stupid CarPlay search thing where it goes, 'What do you want to listen to? Pause. Bing.' And then you talk every single time I want to use the Home View. That's awful. Like this is a thing that clearly has enough processing power. It has enough screen. It's going to be plugged in. So it's not like there's a power issue there.
You're also not driving. You're not distracted, right?
There is nothing preventing this thing from being a computer. Except that Apple doesn't allow it to be. And I guess part of it is the way that it's thinking about switching between users. And if you want to have a bunch of apps logged in to a bunch of different things, like I can see how that model locally would get confusing, but that model's confusing anyway. And if what you're telling me is that this thing is going to learn to seamlessly switch between CarPlay devices, have you ever tried to switch CarPlay devices, Nilay? Like I don't know. This just makes me more concerned than ever that this thing is actually going to work in a way.
We obviously have to get it and review it. We will see. But here's my big question for you is this is a lot of new hardware from Apple coming out, right? The Duo is soon to come out. This thing will come out. There's the new Apple TV. Hot new hardware from Apple. There are rumors today that there will be a touchscreen MacBook from Gurman potentially very soon. This is a lot of new hardware from Apple. It's a big push, right? New CEOs, hardware CEO, here's all the new hardware. And it feels to me like the question we're going to ask at every single turn is did you get the software right? Did you make the right software decisions? And if I was going to be super rude, I would say none of this hardware is particularly like beyond the industry.
The Home View in particular, right, is like it's an LCD display on a previous generation chip.
Sure. But like yes, you could have made this at any point in the past 100 years, Apple.
Sure. I will grant you the Duo has a lot of like engineering in it, but foldables have been around for 5 years. So the wins here are all going to be software wins, ecosystem wins, right? And it's interesting that all of my questions about all these devices are around did we get the software right? Is the ecosystem paying off in the way we want it to or is it actually holding us back? And I truly do not know the answer to that question.
I don't either, but it's been interesting to see like the app developer community has raced to support the Duo. Apple has gone out of its way to help people do this, but also even for a relatively new device that I suspect will not sell in giant numbers just because it's so expensive. App developers are all over this thing and the App Store situation for the Duo on day one is going to look great there. This is what Apple is good at. This is Apple's main advantage over Android in particular. I've been using a lot of different size screen Android devices recently. I had the Alexa tablet. We have Google Books. And I am just struck over and over by the fact that Google is unwilling or unable or both to convince developers to support non-phone devices. Just can't do it. The keyboard stuff doesn't work very well. The large screen stuff doesn't work very well. Apple is good at this. And for Apple to come out with something like the Home View that it ostensibly believes is going to be a big mainstream device. Like Apple gave developers a better story for the Vision Pro than they're giving the Home View. And I'm just confused by this. And I think in the long run it'll probably be like the Apple Watch like you just described, right? It starts as an accessory and slowly over time becomes a computer. But that was a new piece of hardware with tons of new ideas and really really intense constraints. Tiny thing, tiny battery, tiny processor, lots of new stuff going on. This should have none of those problems. Like if ever there was an Apple gadget to just over horsepower into existence with an App Store, it should be the one that is plugged in sitting on your kitchen table. I'm just confused. Like actually the one thing Apple is best at is putting enough processing power and RAM in its computers to make its experiences work.
Like I think there's two reasons people like CarPlay. And I'm like you. I'm like it's fine. It's there. I use it in our cars, but I don't think it's the best thing ever made. But I think there's two reasons people really like it and are very very very vociferous. They'll never buy a car without it. One is just the library of apps. And every time we do like a Decoder CarPlay episode, we get all this email about it. And one always sticks out in my mind. It was a reader who's like, 'I have a Bible app. The maker of the Bible app is never going to write an Android Automotive app and put it in the Play Store for your weird Android car.' Right? There's no way to get that app, that big library of apps. This is like Instagram on the Windows Phone. You can make all the Windows Phone you want. You don't have Instagram, you're dead. There's some long tail of apps that CarPlay supports and that means you have that and every other alternative does not have the one app you want. Right. So, you're back to CarPlay. The second, and I think this is just tremendously important, is that it is using your car's display as just a second monitor. And so, all of the logic, all the application is running on your phone, which has enough processor and enough RAM and a fast cell connection. And so, the computer is good. Like, that's it. And every car maker under specs the processor and RAM and everything else in their car. So you like can't compete with a new iPhone with any car. You just can't do it. Maybe Rivian and Tesla can do it, but I think everybody has the same questions about performance of those systems over time. Apple is like very, they know that this is what they win on. This is why you have a $300 Apple TV. It has more horsepower than any TV app developer can possibly know what to do with except for the one or two game vendors who think they're going to make it happen on there, right? What are you doing with all that power in the Home View? You should just let people run their apps. This all the way back around to should this just be an iPad?
Yes. We got so much response that essentially amounted to, what if Apple had just made a really kick-ass MagSafe mount for the iPad that just worked. I'm like, yeah, that would have been the thing. Just as for the Pixel tablet, you know what I mean? Like, Pixel tablet got everything right. We also missed one detail that a couple of people pointed out that I think is a very big deal, which is that the new Apple TV remote, which is easier to find, is backwards compatible like a long way back in Apple TVs. So, screw the new Apple TV, just buy the new remote.
I think that's what I did with the last Apple TV. They just got me on the remote upgrade cycle.
Remotes and polishing cloth are how Apple really... I feel like I've upgraded my Apple TV remote more often than my iPad.
That might be true. I agree with that. I have a lot of Apple TV remotes now that I think about it. It's like those and lost AirPods are the other thing I upgrade all the time. All right, we should switch gears. We should talk about some AI news. The big news of this week was that Google released Gemini 4 Argon, which is its sort of re-entrance at the frontier of the model race. If you remember for the last, what would you say, three or four months, the scuttlebutt in the AI industry has been that Google fell behind, that with Fable and with Mythos and all of the stuff that OpenAI and Anthropic were doing, that they were way out in front and they had essentially left Google behind. Google then shows up with Gemini 4 Argon, which it says is just as good as those other models. It's also vastly cheaper and it has sort of announced itself back at the front of the race. Robert Hart on our team also wrote a great piece this week pointing out that Meta, which just released a model called Wonderful Full, which is, I don't like Meta.
Well, it's Meta. It's Wonder. It's Muse Wonder. And then there's the big one, the full one, which is Muse Wonderful Full.
Yeah, but they did wonderful. And I don't like that. It's like I just don't like it. Everything about Meta and AI is a little too cute and oddly a little too sexy.
Yeah, it's got like a little bit of an ass on it, you know.
But anyway, what do you make of some of the news of this week? It does feel like there is this sort of renewed sense of real competition in this space for the first time in a minute. Even as all these companies are reckoning with questions about safety and trying to figure out their product stuff, like this who is winning at models race does kind of feel like it's back on. Did you watch Zuckerberg and Alexander Wang on the Cheeky Pint podcast? I mean, everyone watch Zuckerberg drink the Guinness.
I was gonna say we all saw the meme of him enjoying, quote unquote, enjoying a beer. I watched bits and pieces of the interview, but not a ton of it. Why?
I'm fascinated by where Meta is right now. So, I watched most of that interview. As always, it's very fun to watch Zuckerberg talk about communication styles. So, I think they asked him about Wang's like tweets with the butts and Zuckerberg was like, 'Well, you have to be authentic.' And I was like, 'Is that actually your thing?' Butts are who I am. Like, authenticity. All right, whatever. Anyway, there's a part of that where Zuck in particular was talking about the bet they made on training their own model. And Wang was like, 'We need our own reasoning model.' And at one point inside of Meta that was like frowned upon and he was like, 'I think we're going to fail. I think we're making the wrong bet.' And he convinced Zuck to do it. And he like, we're going to write a $200 billion check. We don't know if this is going to pay off. And Zuck was like, but we have to because we are a full stack technology company. We need to control the thing that powers our experiences. And so Meta just in, yeah, in like Zuckerberg fashion was like, we're just going to win. We're just going to do this until it's clear that we shouldn't anymore, right? Like, but this we need to have this technology in order to make Muse work. And the dynamic there that I think is utterly fascinating is they launched the consumer product that people are interested in without the more powerful model. And then now if you pay them money you'll get Muse Wonder and over time I think they'll figure it out. But Meta is a consumer products company. They're good at it. You can feel a lot of ways about Meta. But again I will come back to the Oculus Quest 2 was a great consumer product. They are good at defining a consumer product, marketing it, making it a good consumer experience, and then ruthlessly enshittifying it with Instagram ads, whatever it is. But they're good at it. And I think it's very interesting that their priority was launching the product that was interesting that made investment in the model seem worthwhile.
Right? So, they need a reasoning model because they want Muse to be more capable, but they were happy to be like, Muse is finished, we can give it to people with the capabilities it has today and then over time it will get more capabilities. The same way, I don't know, the iPhone came out and then they were like OLED screens and then the iPhone changed and now they're like folding OLED screens and the iPhone changed, right? But the product like made a promise and kept it. I think Google is releasing all these models left and right and they have no idea what their product should be, right? To the extent that OpenAI and Anthropic have products in market, they've both figured out that their products are enterprise products, right? They are business software companies as you keep calling them and they're more powerful models. You know, I love that everyone's demo case for all these models is like it can use Blender so well and it's like five people in the world need this capability.
Yes. I'm sure they're very enthusiastic, but they haven't made the consumer products. And Meta is the one that has like truly figured out, okay, there's a flywheel here between our consumer product and our massive investment in the models and what that will enable the consumer product to do. And it's great that there's a race on the frontier, but I think that's fun and interesting. It just, man, none of these companies apart from Meta have like put it forward like here's a great, even Dots from OpenAI is enterprise software. I saw Casey talking about how much he loves Dots the other day and he's like I run a business now and I do business stuff with my Dots and I was like very few people are in that position.
Sure, and that's well and good, right? The idea of these things being business software is good and fine and is going to be very successful. I've been reading Kevin Roose's book, The AGI Chronicle. It's really really good. And there is a point he makes very near the beginning that there were kind of two big ideas about AI that are also sort of two big ideas about AI products. And one is do you make narrow AI to solve narrow problems, right? Like do you build and train and refine a model to do English to French translation? You can do it very well, but it is very specific and very narrow. Or do you do the big huge one model to rule them all approach which is much harder and requires vastly more compute but that these researchers found over and over would actually over time outperform the narrow models and he ties this to Google as a company which looked at AI for a really long time as a way to make its existing products better and it's like okay we need this new way of doing translation to improve Google Translate and we can do really great summarization oh that's going to be so useful for Gmail and they looked at these things and they said, 'We have a set of products. How can we poke AI at them?' And you can still see that inside of Google. Even now, Gemini is a product. Google is bad at understanding how to talk about Gemini as a product. And so it just exists kind of outside of itself. While most of its energy is focused on how do we put Gemini into your Gmail and into your Google Docs and how do we beat you over the head with it every time you use one of our products so that we can put AI in there. And in some ways that makes a lot of sense. You get a huge distribution victory. Google has all of these business clients. Gemini 4 is much cheaper than some of the other ones which is a big deal that doesn't get talked about a lot. But like these frontier models are very expensive and Google is running a promotion where they're I think 80% cheaper than the comparative models. Like that's a big deal. But there is no sense to me. I have no belief that Google is even capable of doing what Muse did, which is one entire new thought about how people are going to use AI. Like it's so telling that Google tried to build this thing backwards out of Gmail rather than start with something wholly new.
Oh no. I think it is even worse than that. Google announced Gemini Spark at I/O. They announced the product. They had the other product, the browser use product. I forgot what it was called that they canceled.
And that's what should be inside of Gmail.
Like what do these products, what do agentic products like truly need? They need access to your email. If you were to break it all down, like what is the number one thing they need to be as effective as everyone wants them to be? They want your email.
God, Muse is so thirsty about my... Wait, can I just tell you this is a total diversion. Can I tell you my favorite new thing that's happening in Muse? I play fantasy football. I'm in two leagues and every Monday night I get a report on how the week went and then every Wednesday I get the like waiver report on who picked up and dropped players. So now five days a week maybe I get some kind of update from Muse with fantasy football advice. I never asked it to do this.
It's incredible. It just reaches out and it's like hey here's who I think you should pick up off the waiver wire and I'm like I didn't, you and I have never talked about football.
Have you told it to stop doing that?
No I haven't. I'm sort of enjoying it to be honest.
Okay. Well so Meta is good at consumer products. They're good at engagement bait. They're good at making you come back and use the thing again. They're good at notifications. Like they're good at this. Really, really good at this.
It's like every time talking about AI is so much like talking about a dream you had. Like the stakes are so low for everyone else and for you it's like I almost died. Like no, you didn't. You're fine. You're sitting right here talking about when you were asleep. Anyway, so mine is constantly digging again. I gave it my spam email, so it's always doing the same kind of engagement bait and I told it to stop. I was like, I want you to stop proactively reading my email. I just want updates on the three things that we agreed on, which are mostly like, did I get a class action notice and did you fill it out?
Just make me the like cheapest money you can make.
And it won't stop. And I was like, you have to stop. And I was like, I want you to pay me $200 every time you proactively read my email without my permission.
And it is still trying to figure it out. Like, it is just in the background somewhere at Meta. Some data center is just spinning cycles trying to figure out how to do this. It has not done more proactive email reading. Somebody's like, 'How has this one user used 11 billion tokens?' But he hasn't prompted us in... It's still going. It's still every now and again like, 'Do you figure out this scheme where you're going to stop proactively reading my email?' It's like, 'I'm still figuring out how to make money.'
Let me know when you've done it.
The Roose book is good. It is fun to read. All these people truly hate each other. Kevin had a book party in New York and I saw him speak at it. And he said the book could have been titled That Guy.
That's a really good book title.
Because like they all truly hate each other. And it is very true that Google in particular has the innovator's dilemma. All of its products are huge. They all throw off a lot of cash. They can't disrupt them too much. There isn't a business model for consumer AI yet. They haven't figured it out. And so they're trying to just increment all of those products. And they really should just put Gemini in Gmail and be like, be honest, read your email and send you some notifications about what to do. And I think they're terrified of doing that, right? Whereas the upside for Meta, if they get it right, is sky high. So, we'll see. But that to me is, I'm just looking at, yes, it's great that the frontier models are pushing out there. It's great that like CNBC in the morning now, like the very straight laced morning anchors on CNBC are like RSI is coming and then we'll all be dead.
Weird. It's weird that that has permeated the consciousness because of the capabilities of these models. The products are still not there. And it's...
I think it is worth giving Meta some credit. They figured out a consumer product before anyone else did. And if you look at the Home View, you're kind of like, well, I know what that thing should look like.
Yeah. I have to assume that right now at Google there is either a furious effort to buy Instinct at almost any price or a furious effort to build that exact thing because Google has every single bit of necessary infrastructure to have built an AI bot that works in your messaging app and didn't do it and lost. And like at some point the bidding war for that company is going to be outrageous and I suspect that it's happening as we speak. Two more quick things we should talk about before we get off the AI train. One is, are you caught up on this weird story about the OpenAI employee like planting a bug in one of the conference rooms?
I don't have anything to say about this story. I just think it's wild.
I mean, it's a great story. The part that really got me was he tried to hide it with toilet paper, but tell the rest of the story.
Okay, so as I understand it, basically OpenAI discovered that an employee had planted a bug in one of their conference rooms, ostensibly to record and take trade secrets. This is one of the like important conference rooms that not even everybody has access to. They planted it behind the TV in the USB port, which I think is just a really funny way of just totally dooming the idea of USB ports on television. Like...
You got to get power from somewhere.
It's a really good place to hide a recording device. I'm just saying like I'm not condoning it, but if you're going to put it in a room, that's probably the place to put it. So anyway, this person has been arrested. There's been a bunch of information that has come out about, you know, they booked a one-way plane ticket out of the country. Lots of this story left to tell, I think. But this is also like OpenAI is embroiled in a lawsuit with Apple over the allegation that they're stealing trade secrets from Apple. That is like an official part of their interviewing process. Like the secretiveness and the value of those secrets in this industry I think is just starting to create a huge mass of these stories.
Okay, so first of all, the OpenAI guy, he used his wife's ChatGPT account to find ways to hide a recording device. This is the best fact in this entire story.
I didn't know that. The FBI found that it was his wife's account which was obviously tied to him at some point.
And he was like, 'What are the best ways to hide a tracking mark?' Like incredible facts in this story.
And then he had booked a flight to Saudi Arabia after he planted the device. So he was trying to, a one-way ticket with the family back to Saudi Arabia.
Just very, very good. And then I mentioned the toilet paper. They found a like a thumb drive size recording device and he had for some reason wrapped it in toilet paper and stuck it into the TV.
Yeah. You listen, you're the TV guy. Can you explain this one to me?
It feels worse. I feel like if I walk in and I see a thing in the TV, I assume it's there. If I see a thing in the TV wrapped in toilet paper, now I have questions. I do feel like there's an element of this where he used his wife's chat account and it hallucinated ideas about how to hide a thumb drive and it was like no one will ever notice if you wrap it in toilet paper when what he should have done is gotten a label maker and printed out like E3 and plug, you know what I mean.
Right, like some other thing, like he should have printed like, you know, the iLok dongle.
Or just write the name of the conference room on it and then plug it in and everybody will just assume it's supposed to be there.
Yeah, exactly. Like you just like whatever the conference room, like ours at Vox Media were always named after cities which was very confusing.
So you would just print like Chicago on it and just like plug it in there. It's like fine.
It's like look, it must belong here.
But the number of times at our company when people would be like I'm in New York City and be like yeah I know, like not new information guys, but yeah that's a good story. We'll track that one as it comes. The one other one really quick is, Character AI, which is a company we haven't talked about a ton on this show, but was one of the early what if the AI bot was your best friend/sexual partner kinds of companies. Filed bankruptcy. It turns out in that process was under investigations stemming from a kid who committed suicide not that long after talking with its bot on Character AI. This investigation found or was in the middle of finding some pretty alarming stuff. But now in a really interesting way there are lots of questions, including one of the questions we asked on the site which is what happens to all of this data now, right? Like what happens in a bankruptcy proceeding is it is basically shipped for parts and sold and presumably one of the parts is going to be vast quantities of user data including about children. This is kind of the first one of these that has gone through this at this sort of scale. We also I think 404 Media had a really good story about the finances that go into this and just the unbelievable amounts of money that are being burned in all of these companies. But this is like this is a tiny case study of what I suspect is going to be a lot of companies that burn through a lot of cash and have a lot of user data and are going to just go belly up out of nowhere and what happens after that I don't know that anybody knows yet.
Well, Character AI was kind of doomed in the middle of its story, right? It was started by Noam Shazeer who was one of the early LLM engineers at Google. He was like very frustrated that Google wouldn't do a chatbot. So then he left to do Character AI because he thought everyone was going to fall in love with LLMs. And then Google did a reverse acqui-hire, right? And they brought him back. And so like Character AI is like dead man walking because Google just like hired the team out from under it.
And so like it was always like what are you going to do with this? I think the 404 docs, the bankruptcy docs that they reported on, they were making like a million dollars in revenue. Like it was like nothing. Against all of the spend that they had taken to start the company. So at some point they should have been actively trying to manage what happens to this user data because it was just abundantly clear when Google did the acqui-hire this company was...
Yeah. There's no like when I'm talking about, there's no consumer products like your kids can illegally bang a chatbot is not a consumer product.
No, but these companies are also in a position of they're running so fast and they're burning so much cash and like every day is an existential crisis in so many ways that this will not be the last company that didn't think hard enough about stewarding of that stuff.
And I think a good question is how valuable is this user data? So the other thing that was on my mind is like Spirit Airlines went out of business and there was an auction for their corporate data, right? Because if you have a bunch of data about ticket booking and flight schedules and itinerary and overbooking and fees and just like the operations of an airline, that's really valuable if you're trying to train a model that can do other enterprise tasks. You have a bunch of user data that's like, do you love me? Like we already, they already, they already stole all of Tumblr. Like they don't need more of it, you know? So I think we're going to find that there's like markets for different kinds of data as more and more of these companies go under and we see what happens to more and more of them. But yeah, I mean, should there be a federal privacy law that protects user data when companies like this go out of business? It probably should.
Yeah, agreed. All right, that's enough AI talk for one day. Let's get on to something less controversial, which is the state of social media and X. But first, let's take a quick break. We'll be right back.
Support for the show comes from Jira by Atlassian. AI is only as useful as the context it has to work from. But when that context is scattered across tools, threads, and DMs, that means your team and your AI agents end up flying blind. That's the problem Jira solves. What's the goal tied to your project? What got decided last week in Slack DMs? Atlassian's teamwork graph pulls all the pieces together from Jira, Confluence, GitHub, Slack, and more so nothing falls through the cracks. With Jira, you can actually share the context with agents you already use like Claude, Cursor, and GitHub Copilot. Assign them work directly or connect your agents through MCP. Spend less time digging through endless links and DMs, less time chasing down what got decided and by who, and more time actually shipping with Jira. Learn more at jira.com. That's jira.com.
Support for the show comes from Bitwarden. October is Cybersecurity Awareness Month and here's an upgrade that actually gives you something back: time with fewer password resets, less typing thanks to autofill, and no more trying to remember which variation of your dog's name you used this time. Bitwarden is the trusted open-source solution for passkey, password, and secrets management. It makes secure password management available to everyone with a free forever plan for individuals. That way, all users have a reliable way to create, manage, and autofill strong and unique passwords across accounts and devices. Bitwarden uses end-to-end encryption to protect vault data, while secure autofill helps reduce phishing risk by matching credentials to the right sites, giving users greater peace of mind when logging in. Bitwarden is trusted by over 15 million users and 80,000 businesses across 180 countries and 50 plus languages. Get started with a basic free plan for individuals or protect your business today with a free trial. This Cybersecurity Awareness Month, check out Bitwarden. Visit bitwarden.com to get started.
Support for Vergecast comes from Fetch Pet Insurance. Here's something that might surprise you: every 6 seconds, a pet owner in the US gets hit with a vet bill over $1,000. Be prepared with Fetch Pet Insurance. Fetch is the most complete coverage for dogs and cats. According to consumeradvocate.org, Fetch covers accidents, illnesses, diagnostics, and more. Every vet in the US and Canada is in network. You can build a plan that fits your budget and once approved, claims are paid back in as little as two days. To learn more, go to fetchpet.com/save. That's fetchpet.com/save.
All right, we're back. Nilay, are you on X? I actually, I realize I don't know the answer to this question.
I check it every day now. There's a lot of AI conversation there and I think it is important to see primary sources. Every now and again I will snarkily reply to someone who is racist and remind them I'm from Wisconsin.
That's the thing. That's the thing I do.
I would say that's a reasonably representative use of X. So that's pretty good. So we're going to have this whole conversation. I just want everyone to know that is my experience of X.
Yeah, that's fair. I bring this up because over the last really couple of months now, there's been this slow sort of steady drip of people with some mix of fiery resolve and their tail between their legs going back to X. If you rewind like three years ago, maybe four years ago, kind of in the 12 months after Elon Musk bought Twitter and started to ruin it and turn it into X, a lot of people performatively left the platform, announced they were leaving the platform, renounced their support for this place, and left. I left quietly. This choice was made easily for me because I left on parental leave in December of 2022, which happened to be a really great time to stop tweeting. And I just never posted about it. I just never went back. But I think The Verge has explained in public how we feel about X. But then recently, I think the first one that really came across my radar was MacStories, where John and Federico, the folks at MacStories, decided that they were going to go back to X and start posting. That caused a mini revolt.
Not even a mini revolt, right? They lost a bunch of their staff. People got really mad at them. Full scale revolt. People are very upset. And then just in the last week or so, our friend Casey Newton on Platformer wrote that he is now going to start posting on X again. And then our friend John Gruber wrote about basically that X is the thing again. I don't know John's particular stance on posting on X, but he pointed out basically that X is still the best social network.
No wait, hold on. That's not that sure in of its kind.
No, I well this compare in the universe of social networks in the category of things that look like Twitter.
The title of his post is 'X Once Again Marks the Spot' and he basically says Mastodon is a wasteland, Bluesky lost, Threads is like an empty space filled with people.
But what are those things? Those are all things that look like Twitter. It's like you won the Academy Award for best sound design in a feature film about dinosaurs. Like yes, in that category you might, but best social network is a whole different thing.
Radically different thing.
Sure. Between Twitter and the Twitter clones, there was a everybody believed they were leaving for something else.
Lots of people loudly talked about how they were leaving for something else.
Again, I'm one of those people.
And people are slinking back.
How do you feel about this? I feel like Meta really blew it with Threads. They have a legitimately bigger product with more people and they were so anti-news and anti-realtime feeds that they just lost. What Katie Notopoulos calls it, the gas leak social network. She's a great writer at Business Insider. You go on there and it is just the weirdest algorithm with the weirdest engagement bait. People getting things wrong on purpose, look left and right. Old tweets from like a hundred years ago being rehashed for engagement. Everything that's wrong about Meta is just in your face on Threads.
They have some of the most interesting product design ideas about what a Twitter-like can and should do.
I haven't done this yet, but if you're a podcaster, you can have your podcast card appear between all of your posts on Threads. They have some like real-time NBA chat stuff going on. Like their product ideas are good. Meta is a good consumer product company. Their algorithm ideas are so weighted to engagement that it doesn't matter.
And it is by far bigger than X, by far.
We'll set Bluesky and Mastodon aside for many minutes. We can come back to them. X has decided to win in one category and I think this is why everyone's confused and angry. They have decided to win an AI conversation and tangentially venture capital circle conversation.
If you are a VC and you would like to talk about how smart you are in the company of other VCs, do I have a platform for you? If you are interested in AI and what it can do and being excited about the future of software development, boy, do I have the platform for you. In almost every other case, X is irrelevant. And I was talking to Casey about it because you know he and I shared a revulsion to X at the beginning and his point which I think is valid and I will come to why I think it's valid is if you want to cover the AI industry you have to be in the rooms that they are in and X is that room by far and away that room.
Okay. If you want to cover almost anything else, you might not have to be in that room, right? And so all the people that you are talking about are tech journalists who want to talk about AI and converse about AI and like knock their ideas around and then go home and write about AI or make AI podcasts or whatever it is.
And so all of the principals in the tech industry are on X. That's what they've chosen. And they've chosen it for good reasons, for bad reasons. Maybe John Turnis just doesn't want Elon Musk to yell at him. Maybe they all just hate Zuck more than they hate Elon, right? Like a million, Bluesky is not big enough. Like, fair enough, man. They're making whatever choices they're making. And if you're a reporter, I think it's really important to go look at primary sources.
For better or worse, the primary sources of the conversation are on that platform. This is what I look at every day. I do not feel the need to participate in that conversation. I'm not into giving our work away for free as people know, especially to people who don't value my literal existence. And there are a lot of people on X who do not value my existence. My experience with X is racial slurs. That is just a thing that happens to me when I post on X.
Because people hate Wisconsin.
They hate Wisconsin. I mean, there's like a podcast in X called like The Indian Question. Like we're like 2% of the population. Like we're fine.
We're just all smarter and hotter than you. Okay. I'm sorry. But like that is real and I have a lot of friends who don't see it. They're not experiencing that part of it.
I think that that's happening here. But I do think if you are a tech journalist, you ought to go look at the primary sources. And if your experience on Bluesky is kind of like mine, I started to worry that I was just looking at screenshots of X curated by other people.
Yeah. By people who are mad.
By people who are mad. And the reason I set those aside is I think those platforms should succeed. I think there's some great value in federated social networks and open social networks. And maybe Bluesky's point is never to run Bluesky. It's to enable a bunch of other things to operate. Maybe Mastodon's point is to never really be big. It's just to be the network of things that are on ActivityPub and one day figure out how to interact with Threads, which is just like a holy war. That is all technology.
It is not the consumer experience of where are the important people talking and in this one field in tech they are all talking on X. I understand it. I have a lot of friends besides John and Casey and Ben and all those people who are like, well, I have to go look at it. But here's my counter example and I'm curious what you think about this. It is we are 20 days, 19 days away from the midterm elections. X is irrelevant to the election. It is just not a factor in this election. There was an entire right-wing conspiracy theory about James Talarico being in a mental institution in Ohio. In another world, X would have just horsepowered this into existence.
Right. We've all experienced this in various elections where the wildfire of an X political conspiracy theory became the assignment editor for the entire media. Just didn't happen. Just fizzled. The idea that our candidates are arguing on X or issuing policy positions on X or reacting to X is nowhere in this cycle. And that's the counterpoint.
Okay. I fear you just picked the worst example because the Talarico thing did actually kind of escape containment and it did force the Times to do the piece where they went and looked at his Ring camera records to make sure that he was home and like this one hit, but it's...
I totally disagree with you. It's there. I can show you the coverage. Like the idea that random people on Twitter get to assign the mainstream media what to do remains in existence. It's just there. It's not nearly as often, but it's still there.
But not in a way that breaks through, right? Like in the last presidential election, they're eating the dogs or eating the cats.
Sure. Just concocted out of nothing on X until it landed in a debate and then is now like a rap song on...
I agree with that and I think the thing that is most bizarre about this moment that we're in is that that kind of doesn't happen anywhere. Like in a weird way even as so many other monocultures were breaking, Twitter stayed kind of a monoculture.
So Casey's argument is that Twitter has gotten much smaller, it's much less lucrative than it used to be. You can see its financials as part of SpaceX. SpaceX is a public company now and it's also vastly smaller.
And when you... this is why I reacted so strongly to best social network.
The best social network is like Instagram Reels.
I think there is a strong argument to be made that neither of those things even resemble social networks. But that is where people are having social media experiences.
That's where they're having entertainment experiences. But this is a semantics argument we don't need to have. I take your point and I think in a weird way what actually happened is that we didn't replace Twitter with anything. It all just sort of died. And I'm at a point increasingly where I use Threads more than any of the other ones. I don't really know why. Just so weird.
So, Bluesky. Bluesky is weird because everybody on Bluesky is either like in violent agreement or violent disagreement in a way that I find exhausting in both directions. But everybody's mad all the time.
X I find just like deeply unpleasant. All the people on it, frankly, I find deeply... I just don't have a good time on X and I don't feel any obligation to post there. So, whatever. Threads is like the sort of tamest middle ground for me and that's fine. It's a place to post. People can find it. But there isn't like... I think the idea that you have to go back to X because it is meaningful for most things is true. But the thing about the AI conversation that you're missing is that the AI conversation is the most important. It's the only conversation everybody is having. And that gives X a huge amount of power because...
No, no, they did it on purpose. They made this choice.
For sure, but it's... and for a long time that was politics, right? Like they gave huge weight to political conversations and so they elevated it onto cable news and now it's just AI because of the people who run it.
But the thing is they called that correctly. AI is the most important conversation everyone is having right now. It's the most important conversation political journalists are having right now, right? Like if you can't... it is like you're talking about the people just off-handedly on CNBC talking about how ASI is going to kill us all.
Well, that's their business journal was talking about money.
That's what we do. I think we are at a point now where even that one conversation is the closest thing to a universal conversation being had and it does make X powerful and I think it's a huge bummer that everybody has allowed themselves to stay on X and do that. I think it has a lot to do with the fact that the people on that platform, like they're not there because X is powerful. They're there because their politics have moved in large part the same way that Elon Musk did. The same way that Mark Zuckerberg did, the same way that like... no one is there with open eyes is what I'm saying. And you either do it feeling bad about it or you do it because these are your people and you love it. But it is still powerful in that one way. And that way is very important.
I'll make this distinction. I don't think any normal person should be on X.
I don't think any normal person should be on social media to be absolutely completely honest with you. Like and we'll go with text-based Twitter-clone social media. There is no audience to build there. There is no community to build there because it's also dispersed and most of it is fake. Like I cannot sit here and make one good case for why a regular person should join Twitter, Bluesky, Threads or Mastodon.
Well, as a poster, as a person who like makes things, a person who views things, right? Like those are the most... people are lurkers, right? They're just consuming.
There are better things to view on every other platform.
Or you can just open Instagram Reels.
Read a book. I don't know. But I think the distinction that is getting lost in this conversation going back to X is it's a bunch of reporters.
People who are describing this are reporters.
And I understand why Casey's like, I didn't do the source building on Threads that I was able to on Twitter back in the day.
That is a universal experience that most reporters I know are having.
Because the sources are there.
So you just got to go to the party. You just have to be in the room.
Mike Isaac talked a lot about that, by the way. Mike Isaac, a reporter with The New York Times, talked a lot about basically building his profile as a reporter by being a reply guy in people's DMs.
Oh, yeah. Absolutely. Every reporter in that period did that. We all did it.
My Twitter DMs back in the day were the place to be. That's different now. The reason I'm making Instagram videos is because everyone sees them, right? Like that's just a dynamic that is true. I can talk to my phone camera for two minutes and put it on Instagram and I get a bunch of incoming because even if they're not on Bluesky, they're definitely on Instagram.
Fine. But I don't think that a bunch of reporters saying I have to go engage my sources over here is the same kind of endorsement as this platform has regained its vitality. And I think that's actually the error they're all making.
If you are in tech and you want to talk about AI with other people who are building and using AI, they're all there. That's where you got to go look at it. Reporters should look at primary sources. I believe that with all my heart.
I don't think that means you have to say that like it has regained its cultural power in sports because it has not.
No, it definitely hasn't.
Right. And like there's something there that... this is Casey's argument. It's powerful in its one vertical because it has gotten smaller.
Sure. But all that other stuff didn't go anywhere, did it? Like I do think it's possible that Threads ends up going down as the biggest fumble in this whole space because it...
It just could have done it. And there was a really funny thing this week. Connor Hayes, who leads product at Threads, posted, 'We recently made a change that could increase the posts you see from those you follow.' Like Threads had to make a change to its algorithm so that when you follow somebody you will see the things that they post.
Like what are we doing? How like everybody got so lost in the sauce of wanting to do this like content-based recommendation system from TikTok that they forgot how people want to talk to each other on the internet. And I do think it's possible that Threads may have missed its chance here, but boy did it really actually have a chance to be the thing.
In every other vertical except tech and AI, there's still a chance.
Do you think Meta has a lot of money? They could just get all the sports writers to use Threads.
That's possible. You might be right. But I also wonder like culturally it's now been three years. Maybe we've moved on. Maybe we don't need that thing anymore. I don't know. Look, let me zoom way out. I'm sure people will be mad at me if we're even suggesting that anyone can get any value from X. Fine. Like I can't stop a network effect and I personally do not believe that things go away if you ignore them, right? And the network effect of the tech executives and the people building these things is on X. So if you are like, I'm not going to pay attention to you, the network effect will still be there and they'll still have it. That's just reality. But the mainstream audience is watching video platforms. That is the thing they are doing. And they're watching the video platforms. Like every now and again I get a video and it's just screenshots of X.
And it's like I'm still not on X. I'm just consuming some tweets. That is very powerful. And it's right next to the idea that instead of having a phone, you'll have a little AI agent Tamagotchi.
And I actually think the little video, the scrolly video platform is the center of gravity that will prevent X from ever fully recapturing its magic and prevent the little AI Tamagotchi from ever supplanting the phone.
And like that to me is the real... like sure we can all go look at, we can all go see Greg Jac and John Turnis posting their home view videos on X because that's... they all want to seem cool in front of Elon, whatever it is. But the views for that actually happen on TikTok and Instagram.
Yeah, there's a really interesting argument in that that suggests that scrolly vertical video is a more important format than the smartphone, but I'm not ready to get into that yet. I do just want...
YouTube in that conversation is a good question.
Yeah, fair. I do just want to pick on John Gruber for just one quick second.
Because he writes in this piece, and I should be clear, I think John Gruber is right about most things. I think he's so very wrong about this one thing. He writes, 'The culture on X today is remarkably friendly and inclusive. It is a cheerful place.' That cheerfulness is another differentiator. Is that your experience of X at all?
Again, in the context of like here's a cool new AI product that can do cool new things. I think any excitement to be found is on X.
Sure, that's fair. It is AI people cheerleading other AI people. But that is something...
AI, right? In that there's something there, especially if you work in and around software, you care a lot about software, as John does, that is exciting and fun. There is more tech conversation that is exciting and fun in those platforms than the other Twitter-like networks.
Yeah, that's fair. He also...
My experience of X is nothing like an inclusive experience.
And it is really important to say that out loud.
Yeah. John Gruber and I are both white dudes, which is a really nice thing to be on Twitter. He also says without any fanfare, X has de-Musked its algorithm. And I would say I'm willing to bet that John Gruber muted Elon Musk and forgot about it and now thinks that is the case because that is not my experience of X at all. Do you think people should feel bad going back to X? Like genuinely, what is preventing you from going back and engaging in that in a more real way?
You know these people. Yeah, I mean again I look at it several times a day because it is my job, is our job on this show to go look at primary sources and then talk about them.
The fundamental job of being a reporter is not to report on someone else's reporting but to go try to find some primary sources. And when we do report on other people's reporting, it's almost always people we trust who we know to have done that job, right? Like that's the dynamic.
I think that literally the service we provide here is one, jokes; two, we like each other and people enjoy that; and three, we curate information. Well, like those are I think that's in order what we do on this podcast. And that third thing is really important. So I want to make sure I go look at the primary sources and see the primary sources. I don't think anyone else has that responsibility, but like actually contributing to that platform makes me feel bad and other people can feel differently. Go ahead, do what you got to do. But there's not some part where I'm like, 'Oh, I need to look away from this.' It's like, 'No, my job is to look at primary source information, so I'm going to go seek it out.'
Great. I buy that. But again, where am I putting my social media effort into making videos and I'm confident that is a faster way to reach more people that I've never encountered before than the X algorithm.
It also puts your hair situation much more front and center in the conversation.
Why do you think it's getting crazier with everyone?
I just imagine you like having a hairdresser on the side and you like get your camera up and you're like 'more' and they make it higher and you're like 'more' and they just... that's your Instagram strategy. It's good stuff. All right, we should take one more break and then we're going to come back. It's time for the lightning round. Be right back.
Support for the show comes from IAM by Palo Alto Networks. Every new cloud workload or AI agent your team sends out is another identity that gets permanent access to your systems. And in 2026, those non-human identities with access outnumber actual human identities 109 to one. You need a tool built to manage them all, human, machine, and AI. That's exactly what IAM by Palo Alto Networks was designed to do. It's one identity platform for all and it outshines legacy tools that were only built to manage human employees. How does it work? Well, IAM replaces permanent permissions with dynamic access. So you can lock down every identity without slowing down your own business. With IAM, you get unified control by replacing siloed IAM, identity and access management, PAM, modern privilege access management, and governance tools all with a single control plane. And it does that across human, machine, and AI identities. Secure every identity at your organization with IAM by Palo Alto Networks. Learn more at paloaltonetworks.com/iam. Again, that's paloaltonetworks.com/iam.
Support for the show comes from Outshift, Cisco's incubation engine for frontier technology. You can find multi-agent AI in every enterprise roadmap right now. But here's the problem everyone's facing. Agents can pass messages, but can't think together. These systems silently underperform through cognitive failures, misreadings, unverified claims, false consensus, and conventional monitoring can't detect them. Outshift by Cisco is building the fix with their new comprehensive resource, the Internet of Cognition. It's an open-source guide that's a foundation for building multi-agent systems from design to production. The goal is to give agents shared context, shared memory, and guardrails, all to drive the results we actually expect from AI. You can lean on this resource to understand agentic AI's evolution and how it shifted from isolated agents to orchestrated superintelligence. You can gain insights for how to future-proof your business by tackling complex challenges with multi-agent systems. You can read the paper, experience the demo, grab the code from outshift.com. That's outshift.com.
Support for the show comes from AWS AI. AWS AI is powering transformation across the world's leading organizations. Moving AI from possibility to production at scale, like how AWS AI helps Formula 1 turn race day data into real-time insights that bring fans closer to the action. Or how it helps Booking.com personalize every trip for travelers around the world. AWS AI is even transforming how things get built, helping Siemens power the production of smarter cars across industries. AWS AI is helping businesses innovate faster, lead with confidence, and transform what's possible today. AWS AI is how.
All right, we're back. I'm not even going to ask. I just... this one I know. Brendan broke containment in my world this week. It's time once again for America's favorite podcast within a podcast. Brendan Carr is a dummy.
