The Vergecast - How much should you tell your AI agent?

Episode Date: October 6, 2026

The next big thing in AI, at least if you ask the tech companies, is assistant agents. David and The Verge's Allison Johnson have both been testing products like Meta's Muse, OpenAI's Dots, and startups like Instinct, in an effort to find out how much you can really get done by messaging with an AI tool – and whether the productivity is worth the tradeoffs. Further reading: Paramount and Warner Bros. Discovery complete $110 billion media megamerger Amazon reportedly pays $120 million to stream the Emmys for free New leaks reveal more details about the Fitbit Edge including its price A Trump Mobile breach may have exposed data of more than 3,600 people OpenAI’s Dot agent can help you work and order dinner Meta’s Muse AI agent is an effective helper and great at spending money The Vergecast has been nominated for a Signal Award! You can vote to help us win here. Subscribe to The Verge for unlimited access to theverge.com, subscriber-exclusive newsletters, and our ad-free podcast feed. We love hearing from you! Email your questions and thoughts to [e-mail] or call us at 866-VERGE11. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Hello and welcome to the Vergecast, flagship podcasts of secure password vaults in your text messaging. I'm your friend David Pierce, and today on the show, we're going to talk about agentic AI messaging systems. So there's been all these new products, right? Meta's muse, open AIs, dots, instinct, a bunch of other startups coming out all with the same idea that basically what you want to do is message with an AI assistant that will then use agents to go and accomplish stuff on your behalf. And I say stuff in the loosest possible way because they all have big, different ideas about what you might want your agents to do. Alison Johnson on our team has been testing all of these. They're all very similar in certain ways and all very different, in interesting and important ways as we try to figure out how people actually want to use AI.

So she's going to come on the show and we're going to talk through how we feel using these products, whether they're any good and which one you should be using. All that is coming in just a sec, but first, here's everything else happening on the verge today. I'm Travis Larch, and this is 90 seconds on the verge for Tuesday, October 6, 2026. The $110 billion mega merger of Paramount and Warner Brothers Discovery is complete. The combined company is called Skydance, and it includes Adult Swim, Alloy Entertainment, Animal Planet, BET, Bletreiberine, Cartier Network, CBS, CMT, CNN, Comedy Central Cooking Channel, DC Comics, Discovery, Food Network, Golf Digest, HBO HG TV, Magnolia Network, MTV, Magnolia network, MTV, Network 10,

New Line Cinema, Nickelodeon, Plytoe, Poet TV, Quest, Science Channel, See It Now, Studio, Shed Media, Spusonian Channel, TBSC, TNT, TNT, Travel, Channel, Tour Classic Movies, TV, Warner Brothers. Also, Amazon bought the rights to stream the Emmys, so starting in 2027, you'll need to go to Prime Video to find out who wins among however many networks we have left at that point. More info keeps trickling out about the upcoming Fitbit Edge. The new leaked details shared by Deal Labs suggest that it'll end up costing around $180 in the U.S. and that it'll have an always-on OLED touchscreen and offer up to seven days of battery life so long as you don't leave the always on, OLED touchscreen always on. And finally, PC magazine reports that Trump Mobile

appears to have suffered a data breach. The data set includes 3,615 people's names and contact information. I have nothing to add other than 3,615 is an awfully specific number. You can read more at the verge. That's 90 seconds on The Verge for Tuesday, October 6, 2026. Visit BetMGM casino and check out the newest exclusive. The Price is Right Fortune Pick. BetMGM and GameSense remind you to play responsibly. 19 plus to wager. Ontario only.

Please play responsibly. If you have questions or concerns about your gambling or someone close to you, please contact connects Ontario at 1-866-531-2,600 to speak to an advisor. Free of charge. BetMGM operates pursuant to an operating agreement with Eye Gaming Ontario. New from Nespresso. Blend wellness into your coffee routine with the coffee plus range, infused with functional benefits. Choose the coffee you love with added B vitamins, like coffee plus B12 to help support immune function,

and coffee plus B6 to keep your day moving. Or go with the flow and choose ginseng delight. Our new double espresso with ginseng extract. Whatever lies ahead, don't change your morning. Let your morning change you. Discover coffee plus on nispresso.com. 50 years, Canadians have wondered what the names of IKEA furniture mean. Like Natopa, the guardrail for children's beds.

Natopa means giving your favorite person room to grow. It means their safety and your piece of... No. In Swedish, Natapa means night monkey. Mom, he's back. IKEA, 50 years of Swedishness at home. Look, he's wearing little pajamas. So cute.

Let's talk agents. The Virges Allison Johnson is here. Hi, Allison. Hello. So are you the kind of person who outside of journalistic responsibility would have downloaded an app like Muse and tried to find things to do with it? Boy is that hard to know because I was told immediately that I would be using these things. I think it appeals to me in the way that there are too many things to do and not enough time. So, yeah, maybe. I guess maybe another way of asking this question was how hard was it for you to figure out things to do with all of these agent assistants? Because I think I've been gone for a couple of months and all I have had time to do is sit and do things that only require one hand while I have a baby in the other hand. And one great thing to do is just goof around with weird AI assistants that claim to do things for me. And so in a funny way, I've been the exact optimal. target for these things in that I have stuff I need to accomplish and I need the simplest

possible way to accomplish them, including not having to type in hexadecimal passwords all the time. And I understand that I am at like a very particular moment that is not representative of how most people experience these things. So what is it like for you, the sort of cold start problem of like you've tested a bunch of these things? Do you open it up and immediately have a bunch of ideas in your head about how to use a product like this? It is a little bit of the blank page kind of cold start. artifact where you're like, huh, what do I do? What is my life? Like, do I have to? And then, I don't know. I

was coming off of, like, an extremely busy week, and I've been traveling. So I sat down with Muse, and it didn't take too long. I was like, okay, I do have a backlog of stuff that is just piled up. Like, my husband's been taking care of the kid alone for a week. He hasn't had time to deal with stuff. I have not had time to deal with stuff. So there was a list of of things that I could get it to do. So that helps. Yeah. I do think one of the things I always see on my social feeds, which I always enjoy, is the people who defend AI are constantly like, if you think people don't have giant endless tiny low stakes to do lists all the time, you have never met a working parent.

And like, while that obviously doesn't describe everybody, boy do I agree with that thesis. So you've tested recently, you've tested Meta's Muse, you've tested OpenAIs, Dots, and you've tested this app Instinct, which is a startup that's running mostly through iMessage. I'm fascinated by your experience with those three in particular because I kind of think you've seen the spectrum of the ways this thing can work, right? Everybody, I mean, everybody is building this AI assistant. everybody is building a thing that you can message with that will go to things on your behalf.

This is the only idea in Silicon Valley right now. The three flavors of it, as far as I can tell, are the three that you've tried. There's the one that lives in your existing messaging app. There's the one that just is a messaging app, which is instinct and muse, respectively. And then there's dots, which is a feature that lives inside of a larger AI ecosystem, right? It's not dots in an app. It's dots inside of chat GPT. of those three dedicated app inside your messaging app part of a broader AI thing which feels right to you so far which have you gravitated to the most

they've all kind of like shown themselves to have different strengths to me but the one that strikes me as like the most useful in a day-to-day way and the least intrusive is instinct because it is just you in your messaging app and you just, that makes it kind of everywhere and nowhere at the same time. Like I'll think of something and I have my smartwatch on. So I'm like, oh, I can fire off a text to instinct or you're in the car and you can do, you know, car play kind of things. So, and it's also the most like just to the point getting things done. Like you're kind of constrained in that.

you're typing out a little message on your little smartphone keyboard. So you don't get long-witted responses. It's very like got it doing that thing or even just an emoji reaction, which is funny. I'm like, don't pretend you're a person with like a sense of humor. I don't know. Yeah, anytime it gives me the like thumbs up tap back emoji. I'm like, you don't have thumbs. Get out of here.

Yeah, come on. What do you know about thumbs? You don't know about thumbs. I think I agree. And part of the reason I've been thinking about this is I will say for like regular people, I'm all the way out on the Dots approach, right? And I think your experience with Dots was that this is, I think you called it work software that will order me a burrito. And I think that's exactly right. And I think that's very useful as far as it goes for people who will use it for business. I think that's a non-starter for regular people in their regular lives. Mews, I think, works because it feels like a messaging app. And obviously, meta is going to put. muse everywhere. It's going to go in WhatsApp. It's going to go in Instagram. It's going to go. Like, meta will eventually percolate this to lots of messaging apps. And I imagine that's where it will be very successful. The place where I'm torn is instinct is very understandable in the way that you just described. Right. I know what it is because it's a thing I text message. It doesn't have any

pretense of being a thing that I've installed on my device. It doesn't try to ask for any kind of disc access. I know the box around instinct in a way that's really helpful. That also causes a bunch of problems. Like, have you ever had to open up the instinct browser when it's trying to do something and it gets hung up on one of those, prove you're a human moments? Have you ever had this happen? Not with instincts, but I did with dots. And it felt, it felt weird.

I felt like I was like an underage kid would ask you to buy cigarettes or something, you know? So for whatever reason, for me, with instinct, it runs into cloud flares, prove you're a human thing all the time. And when you open up the browser, it opens up the browser in the little thing and you have to then like mouse around and tap. And it just like that breaks, right? Because it has to exist in these very specific confines. Yeah.

But then with Muse, you get access to all that stuff. And right, it can do more things that way. It can more accurately connect to the connectors because it can just offer you a list of connectors that you can tap on. So like, Muse is very simple in that way, but it's also more. complicated. And I don't know, I've really gone back and forth between, I have had that same experience that you have where it's really nice to just fire off a text message and not really think

about any of the infrastructure around all of it. But to the extent that you kind of have to have some infrastructure, like you have to do a little bit of work to let it access your Gmail. It just is a little bit easier that way. I don't know. I've struggled with this. Mews has felt more like if I'm sitting at my computer, which is funny, like, you know, as a meta product, but it was more useful for me when I was sitting down. I didn't even necessarily have the desktop app open. I was like in a browser window. But then you have like more canvas.

You can see its little computer that it's clicking around on. You just have that like quick, clear view into like, okay, I have my mouse controls to take over when I need to do that. Yeah, instinct by its nature, I think kind of like abstracts that away. And it's just sort of like, don't worry about it. Let me handle this. I got it. You sometimes get that like secure login link to, you know, get your Gmail or whatever credentials stored safely, et cetera.

We love it. You run into that sometimes and you're like, oh, this is, this is software. You're like, I see you behind the curtain. There is an app running here. You're just not showing it to me. Yeah, right. But otherwise, it's like, you just text messaging, a very helpful little guy somewhere, I guess. Yeah, I've really come around on it on that front because I think in the early days of some of these apps when most people were using it for, you know, clawed code and open claw and stuff.

I was like, the idea of accessing this stuff through Telegram just feels wrong. Like I need a different kind of interface than just messaging. But for the stuff that these things do well, which is very basic. one or two step tasks, just send a text message actually kind of works for me. And I've really come around on the idea that actually maybe I do want to message

some AI system to accomplish things for me. And maybe that's good enough. What's the most surprising thing that has worked for you in testing these things? Either something that you really pushed that worked and you didn't expect it to or something that worked in a way that you didn't mean it to but was cool anyway.

Like what stuck out is kind of the hero experience so far. Oh, gosh, several things. First of all, Mews as a shopping platform, just so sleek and frictionless. It's like you get your credit card in there one time through like Stripe or whatever. And it's like, I'm just going to run all over the internet

and buy things for you. Somewhere an army of Facebook ad product managers just got a race. Just when the minute this published. hats off. Like, it's just like, you don't even have to click a button. You just type the words, go ahead. And it just fires off the transaction for you. This is like problematic in many ways for me personally in my bank account. But also just kind of like, yeah, the product working as intended is almost like too good for those purposes. I think the other thing that kind of caught me off guard

unlike the other side of spectrum was working with dots, dot, you know, my personal dot, which I called Dotty. Dot has dots. I think it's dot and dots. Dot is, plural of dot is dot and also dots. Could also have dots. Yeah. My dot is named Doddy McDot face because I can't help myself. But I was trying to do all the like consumer-ish stuff. I was like, I'm trying to set up a new ISP at my house. And it was kind of clunky and, like, not working great for all that stuff. And I was like, oh, you just got to do work. Like, this is just work software.

So I started, like, bossing it around. And I was like, go update my personal website and kind of came back with some stuff. And it has the option, I think, a few of these do. Maybe they all do where you can, like, call it, quote, unquote, and, like, talk to it. This is a big feature of Dot in particular, though. the voice mode assistant bit of it all. Maybe it's sort of like more present in a way where it's like, you can just, you know, just call your dot.

I was like, that's going to feel weird and dumb and I don't want to do that. And I called it. And I was like firing off just a bunch of stuff I wanted it to do. Like it gave me a draft like kind of proposal of what this website redesign would look like. I was like, I'm just going to talk to this thing for 10 minutes. say everything that's wrong with it and just kind of went like section by section. And I kept being like, so you're with me? You got all this. It's like, oh, yeah, yeah, good. And then you just sort of like end the thing by being like, okay, couldn't you go do all that? And it was like,

yep, got it. And it did. Like, it did a pretty good job. We didn't, it wasn't perfect. But I was like, that actually was better than sitting and typing, you know, paragraphs of thoughts about what I wanted. So it was surprising in that, like, I was okay with it. And I did not think I wanted to talk to the robot on the phone. That I have really struggled to adopt in my own behavior, actually, because I think I'm so used to commanding computers that my instinct, even when using one of these tools, is to be very short and very direct and very sort of turn by turn. Like, do this thing that I'm going to explain very clearly and in one sentence so that you don't get distracted. You're going to go do that thing and then I'm going to give you another sentence of

things to do. And actually, there seems to be a lot of evidence that the best thing you can do is just ramble nonsensically at it for like 10 minutes. Yeah. And that actually these tools are very good at distilling that into something more useful than you would have commanded it in the first place. It's really interesting. I just, I am so bad at remembering to do that in using these tools. It makes more sense, I think, in work. Like, you giving it a large data set, which is just me rambling, as opposed to like, I'm hungry, get me a burrito, you know, is like, not quite the same his case. Yeah, although I don't know.

I find myself increasingly far apart from our commenters on how I feel about a lot of these things. I think my moral feelings about AI, I largely agree with our commenters. I think AI is built on bad things. I think it is run by companies that are doing bad things. the world. Like, I think it's all bad. I think these products are actually really good and useful in a lot of ways. And I think the cognitive dissonance of that is like the great question of our time. But right now, I think like I look at the DoorDash news from last week, right? Where they're like,

you can just, you can just chat to DoorDash and say, order me the thing that I got. I want that so bad in my life. Like once, once a week, I'm lying, four times a week, I go into the Toast app and I order the same thing from a coffee shop and I go and pick it up. And if I could turn that from 10 taps in toast to just order the usual in a text message, that's good. That is a better user experience for me than being in a toast tap. And this runs into, I think, the thing that you have been reckoning with a lot and I have been thinking about a lot while, you know, sitting on my couch on leave, is this question of is, is that friction by definition a good thing, right? Like you look at the comments on any piece that you've written about this stuff.

And you've talked about this on the show, too. All of the responses are, why are you trying to automate being a person out of your life, right? And I think David Emel and Liz Lapato and V Song did a great episode talking about this stuff. Like, what does it mean to be a person is strangely important to being a technology journalist in 2026 in a way that I didn't? Like, basic humanism is at stake right now. And I feel like for you in particular, you've had to really think about this in sort of a brass tax way. Like what does it mean that I am giving these tasks and not these tasks away? Again, I think you are required by the virtues of your job to overindex on doing the thing.

Like your own morals get sort of subjugated to what we need to do in order to thoroughly test these products. But how are you feeling about that? Like, do you spend a lot of time with Muse and feel sort of, icky in the way that, like, I feel when I look at TikTok for too long. God, I have a lot of thoughts about this. And it sort of started to percolate when Google was adding the, like, automation stuff to Gemini, like, way earlier this year where it would, it would, like, tap around on the DoorDash app and order some food for you. Sure. And you see it work once, and it's sort of like, oh, yeah, of course, like, why wouldn't I do this version of just saying something to it?

and going, hey, order that thing for me, rather than tapping around of like all things being equal, I pick that one, you know? But there's... You know, it's interesting. There was a, sorry, I don't mean to interrupt, but there was a really great piece in the Atlantic a week or so ago that made an argument I have not been able to stop thinking about that basically what we have done with technology is convenience our way out of ever interacting with people, right?

And so instead of walking into a place full of people and talking. to the person at the counter and having actual human interactions, we've all just reduced ourselves to I can do all the work on the screen. I find that argument incredibly compelling. Like, if you want to say throw your phone in the ocean, go talk to the barista, I'm here for that argument. I have a much harder time with the using Muse is a morally inferior tool than going into the Toast app to order to go. Do you know what I mean? Yeah. I think so much of our lives have already been kind of commoditized and automated anyway. And it's not like an endorsement of how we live, but it's like, yeah, like I do place a to go order with Starbucks as I'm running out

the door. It means I don't talk to the barista. I've had very lovely conversations with the barista at the Starbucks I go to. One of them like saw me come in throughout my pregnancy and I came in. She was like, did you have that baby? I was like, yeah. Yeah, that's so nice. You know, you have those moments. That's the stuff right there, yeah. That's a spice of life right there. But it's sort of like when you have the option for convenience and you live a life at the speed of, oh, my God, we got to get to this kid's birthday party.

We're late for it already. And we just finished up the grocery shop. You know, there are like your life has already accumulated so many tasks and things. You're like, yeah, I would. will automate this one thing and, you know, a coffee at Starbucks is a luxury and a convenience in itself. But, yeah, God, I think about this a lot. It's sort of like, you know, I think about my dishwasher. I love my dishwasher. I will not live without a dishwasher, you know, if I can help it. Would I gain something, I don't know, in sort of like morally by doing my

own dishes, you know, after every meal, having that time to totally think, you know, about anything. I don't know. Maybe, maybe I'm missing out on some kind of like spice of life experience. But you live in it, like, I just keep coming back to we live in a society, you know, and it's like we have, all of our coworkers and bosses have dishwashers. And if I was the one insisting on, And, you know, now I got to slow down and do all my dishes and my friends. You know, I don't know. It's kind of a strange example. But I think we live this out in a lot of little ways of like if I, if I insisted on, I don't use email.

If I was like, I'm, I use pen and paper and I'm going to talk to my colleagues that way. Like that. I call that Christopher Nolan. If you decided to Christopher Nolan. Yeah. Or like Bill Murray can get away with the, he's like, yes. Like very elusive and will only, you know, contact you on his terms.

It just appears at softball games and that's how you get in touch with Bill Murray. It's like, I think that works for Bill Murray. I don't think I have the same clout to be like I will not be emailed. So we're kind of, we have to go along with the game. And I think there's, there's a danger of, I don't know, things keep speeding up. It gets easier to put more. tasks on your plate. So then you put more things on and you've got to automate more stuff. It's, it's, it gives me a headache like every single day. I agree. And I think that is so much of

the tension so many people feel with AI right now. Right. There is this sense of even if I don't want it, if everyone else uses it, I will be left behind in some meaningful way. Right. And it's like, you see all these things about people who don't want to use it, but are using it every single day. work. And it's will be, it's like, well, I don't want to get fired. Like, at some point, if I don't do this, I fear getting fired or if you're not getting a job out of college, so I have to use these tools in college, even though they make me feel bad. I think that thing you just described where like, well, everybody else assumes I have a dishwasher is actually real and important and a complicated piece of this. And I think to some extent, it's one of the reasons I like tools like

Muse and Instinct, because they don't then require you to jump through a bunch of hoops to access some of this stuff. It is like if we are going to let this technology exist. And I mean that sincerely as an if. And it is a question we should continue to ask at all levels of government and society for a long time. But if we're going to let it exist, it has to be this easy. And I think something like this is the right kind of easy in a way that like I've seen people in my life who never would have touched Claude Code with a 10 foot pole using Muse and finding actually useful things out of it. And I think that's good. I have a hard time being against that, even given all of my complicated feelings about AI. Right. Yeah. I think tool is the key word. And I find myself

using a mix of stuff. It's like there's stuff I'm going to go to Siri for. There's stuff that I would, you know, throw to an instinct or amuse. And it's, they are all tools in the toolkit. And none of them has been my personal AI assistant, you know, for everything. Support for the show comes from Framer. If you want to create a website that looks great, feels handcrafted, and comes together fast and seamlessly, then look no further than Framer. It's a complete website platform, not just a builder. So teams can launch and keep improving their sites in one place.

It helps creators, teams, and businesses ship production-ready sites faster than ever while getting every detail right. You're able to prompt, inspect, edit, and publish in one place at a whole new pace. On Framer, agents and humans work in tandem. Agents bring speed and scale, while people bring taste, judgment, and control. And that all takes place in the same workspace where the real site is designed, managed, reviewed, and published. Thousands of businesses from early stage startups to Fortune 500s or choosing to build their websites in Framer, where changes take minutes instead of days. Learn how you can get more out of your site from a Framer specialist or get started building for free today at Framer.com slash Verge for 30% off a Framer pro

annual plan. That's Framer.com slash verge. Rules and restrictions may apply. Support for the show comes from Shopify. Starting a business requires you to be resourceful. And the more you flex that muscle, the more you start to see opportunities for growth. For example, launching your business with Black Friday approaching means you can take advantage

of the extra online traffic. Shopify is there to help you make it all happen. Shopify is the platform for a record-breaking, reliable holiday selling season for businesses at every stage. Their pre-built website themes and AI site builder make it easy to launch a beautiful,

professional looking store. And you don't ever need to touch a line of code to do it. They also have a mobile app that lets you run your store from anywhere. Their AI e-commerce expert, sidekick, can help you manage almost any business tasks with a single prompt, including holiday discounts. With Shopify, you can easily sell everywhere people shop, online, in person, across social media, and even in AI chats. If you're ready to launch your business before Black Friday, go to Shopify,

dot com slash decoder to start your free trial. That's Shopify.com slash decoder. Go to Shopify.com slash decoder today to hear your first. Support for the show comes from Satelli. Every dog owner knows that sinking feeling, the one you get in the pit of your stomach when suddenly you realize that you don't know where your dog is. The Satelli Collar Go is a GPS dog tracker that helps keep tabs on your dog

with real-time GPS tracking, escape alerts, and intelligent monitoring. Real-time GPS location tracking and virtual boundaries can help keep your pet out of danger while providing freedom. If you live on a large piece of land or in a rural location, or even just like to let your dog run off leash every once in a while, the satellite collar go can give you peace of mind. But the satellite collar go isn't just about tracking where your dog is. It's also about tracking their health. Satellite Sense AI helps you spot changes. in your pets health, particularly when you aren't there to watch.

With Satellite Collar Go, you can know where your dog is, what your dog's doing, and get insights into how your dog's feeling. Go to Satelli.com to get 35% off with promo code Vergecast. That's S-A-T-T-E-L-A-I.com to get 35% off with the promo code Vergecast. Satelli.com, promo code Vergecast. Visit BetMGM casino and check out the newest exclusive. The Price is Right Fortune Pick. BetMGM and GameSense remind you to play responsibly.

19 plus to wager. Ontario only. Please play responsibly. If you have questions or concerns about your gambling or someone close to you, please contact Connects Ontario at 1-866-531-2,600 to speak to an advisor. Free of charge. BetMGM operates pursuant to an operating agreement with Eye Gaming Ontario. Have you found yourself not doing things with AI on purpose?

Like is there a line in your mind where you're like, well, this is a task I really ought to do myself or I don't trust the AI to do this for some reason. Have you found that line? I kind of like try to challenge myself. I caught myself being really like over-eager to have, you know, whatever chat assistant is on my phone at the time. Just look something up for me that I can't remember. where you're like, something's on the tip of your tongue. You're like, what's the name of that thing that we saw? What's the band?

The blah, blah, blah, whatever. Those moments, I'm like, no, I'm going to work for this. And it's not a new problem we've had. No, people have been saying that since, like, Google was invented. Yeah, right. Oh, no, Google, nobody has to remember movie names. Yeah, we're all going to be so dumb. Like, and there's something about it being as frictionless as I just yell at my phone.

in my phone is, you know, like across the table. I caught myself doing that more than once. I was like, you know, I could have come up with that if I had thought about it for more than two seconds. But that's like a little, a weird little gem of like, I'm just going to have this moment and really try. And maybe I'll come up with it. Maybe I'll just have to keep wondering for the rest of the day. I don't know. What about you?

I think I agree with that. I have definitely caught myself looking for reasons. to do things myself in an interesting way. Like, I'll just give you one example. So my wife and I just had our third kid. And the immediate discovery upon having a third kid was that all of our cars are too small. We have two of them and they are both too small.

Right now we have three car seats and our oldest sits in the middle and can literally like hit both of his siblings in the face simultaneously, which he does many times a car ride. Yeah. And so we're like, okay, we need a car with three rows. I am pro minivan. Anna, my wife is just not ready to be a minivan mom.

that's okay and I accept her the way that she is. So we're in the middle of shopping for a bunch of cars. So I went to, I think it was Mews that I did this with. And I was like, okay, here are my requirements. I need a car with three rows. I need a car with captain seats in the second row so that you can access the third row more easily. And I need it to be, I think I said, under 50,000 miles, ideally a hybrid, but price over everything. And ever since then on Wednesdays and Saturdays, it has sent me just a bunch of car listings that might apply to me. And I had this moment of being like, this is useful in the sense that it's like sending me listings to cars that I maybe should buy. And at one point, it asked me a question that was literally like, do you want me to buy this one?

And I responded yes to a previous message and it starts loading. And I was like, oh my God, did you just buy me a car? It didn't. Thank God. But it was one of those things where I was like, okay, not only is this automating this process. this is actually a lot of work I need to do myself. I need to, A, go see what other cars are available other than the two that is identified for me. I need to go test drive those cars to see if they work.

I need to spend a lot more time thinking about how much I care about like a new car versus use car. There's just like, this is a thing that actually requires a lot of work from me to work for me. And I cannot offload this. Even though it actually, it successfully accomplished the task that I gave it. And it was completely the wrong tool for that task. And that's just, I've thought about that a lot. for the last couple of weeks. Yeah.

But it was the, it was the, it almost bought it for me thing that was actually the most alarming moment that I've had because it could have. It had my credit card information. My credit card against all rational economic, you know, incentives has a high enough limit that it could have just bought me a car. Yeah. It shouldn't have, and I would have been homeless very quickly, but it could have just bought me a car on my credit card.

Oh, but that would have been such a good blog. It would have been my favorite expense report of all time. if I had just bought a car from Muse. But it made me think about the level of permissions I'm willing to give these tools. And again, I think you and I are both incentivized by our jobs to overindex on give it more information. And I think we've both been turned into a little bit of nihilists about this where it's like, well, I've already connected everything else to my Gmail. Like, what's one more? But I think, like, I open the Muse app and you go to the connectors page.

And it's just every bit of personal information you could possibly ask for. It's your email. It's your calendar. It's your health data. It's your financial data. It's your medical records. It's everything. And the thing that has most broken my brain is that every single one of those you turn on and connect makes the product better.

In a direct linear way, the more data you give these tools, the better the tools are. They will do more stuff for you. They will make it more personalized. They will be more successful. It is just better in every way. And then on the one hand, I think you're completely insane to give all of that information to meta. Meta in particular, all the promises it will make you, you should not believe meta. I don't know how to feel any way other than that.

Meta's entire history suggests it is not a good steward of that data. Turn most of them on anyway, because it made me use better. Where is your head on this stuff right now? They want your passwords. They want your credit card information. They want access to all of the other apps that you use for everything. How are you feeling about turning all that stuff on in all these different apps? I feel, yeah, probably I'm a little more liberal than I should be with the passwords.

You give me like a secure login vault and I'm like, okay, I feel pretty good about that. Well, you use the word vault, so it's probably fine. Yeah, right. Maybe I'm desensitized in the way that like I live by my password manager and it still kind of boggles my mind when I'm talking. talking about password managers to people who are like, but then the bad guys could just steal, they could log into your password manager and steal everything. And I'm like, no, no, no, no, no. It's not like, it's way better. But also like, buddy, if you get the password to my password manager, I'm long dead. You know what I mean? Like, go with God. Yeah, yeah. I just got to, I got to move to the woods if that happens. I don't have anything left. But I think, so I'm, I'm constantly logging into new stuff on every phone that I test. Maybe that has desensitized me a little bit where I'm like, what's another password manager thing? Like, sure, have Adam.

And they want you to feel that way. To be clear, all of these companies and all of these services are trying their hardest to make sure that we all feel that way. And it's really hard not to. Yep. And it has worked. I'm like, every website I bought anything from in the past 10 years has some version of my credit card information. And that gets hacked once a year anyway. way for, you know, whatever reason. So yeah, there's a little bit of like, oh, what's one more?

What would you tell other people, though? Like, if, I'm sure you've had the same thing I have where, like, I think Muse in particular has so sort of escaped containment that, again, like, people I would never talk tech about are asking me about Muse. I'm sure you've gotten the same thing. Yeah. What do you tell those people? What's the right reference framework there? You got to like listen to your gut on it. I think if you practice good password hygiene generally and you give it access to something like your email, it feels a little bit like a moat of, well, it can't just run roughshod over every account that you have because it's all it needs is your email to get a two-factor authentication to get in something else. I don't know. I think. think my advice is like, yeah, think about it, be sure you know what you're going to get out of it and how that's going to benefit you. And like, yeah, just do the right thing with your passwords to begin with. And hopefully that helps you when the robots decide to kill us all in 10 years. Yeah. The only, I think, Pat advice I've come to on that front is only give it one credit.

card number and keep careful watch of that credit card. Yeah. That's just like no matter how many cards you have, whatever, give it one number and check that account every damn day. Yeah. Because like there just have been enough stories already of weird things happening to people that like I don't think meta has any particular kind of malice. I don't like I don't think this is a sciop to steal any your credit card information from

meta. If meta wants to get your credit card information, it's like it's readily available. Yeah, yeah. But there is this increasing set of evidence that suggests that these models and these tools will do things that are not predictable and that we don't expect and sometimes don't ask for. And I don't know if this has been your experience at least, but meta at least is super thirsty. Like, Mews is constantly like, hey, David, here's a thing I found. Hey, David, do you want to buy this thing? David, David, it's in your calendar.

Did you know that this is in your calendar? And I'm like, Mew's, yeah, I put this in my calendar 15 minutes ago. Right. Yes, I'm aware that this is in my calendar. Thank you very much. And you can just see, again, meta in particular is really good at selling you stuff. And you can just see how much they want to shrink that line. And so I'm like, if you want to give your credit card to Mews, I do not blame you.

I also think you should look at your statement every single day. Yeah, because you might buy a car on accident. You might buy a car by accident. Don't be like me, people. Only buy cars on purpose. That's a pretty good rule to live by. Exactly. So you've tested all of these now.

What do you think is going to stick? Like in your sort of day-to-day life, are you still finding yourself opening news, even having written about it to do something? What behavior still lingers? I haven't had, I haven't gone back to muse. I had, I set up instinct to do a thing every morning where it goes through my personal email and gives me the highlights. I'm like, just only tell me the things I got to pay attention to. And that has been like pretty useful. So I'm like, oh, right, I'm going to, I'm going to.

going to keep that checked on for now. Yeah, if I come back to some kind of like enterprise software need, you know, I can see like having Dot do more stuff for me in that way of like, I was like, why don't you go out once a month and find some like professional development opportunities? I'm like terrible about seeking those things out. I'm sure they're out there. And it came back with I was like, okay, pretty good. Like, those sort of things where it's maybe a regular cadence and it's something is like, I know I'm going to forget about this. It would be good for me if I had this information. Can you go find it? It's a good idea. Like, yeah, those kinds of use cases, I'm like, I can live with this. Yeah. That's a very like souped up Google alert thing, right? Like, go go find information that is relevant to me about this thing or at this time. I think that's useful. And that seems to be a pretty sticky thing for a lot of people. We've seen this from Google did this with the CC bot that it's

marketing is like for families, but it'll go through your email and be like, here's your itinerary for the day. I've heard from people who really like that. I get that email every morning and it does nothing for me because all of my stuff is in work accounts. Yeah. That doesn't work. But the eternal Google problem of

people have multiple Google accounts. But there is something to that basic just like there's a lot of information out there. Bring it to me sometimes that I think is actually pretty useful and I think these tools are really good at. Yeah. Yeah. All right, Alison, thank you.

It's nice to see you again. Thank you for dealing with all of my feelings. And as you test more of these things, keep me posted. I'm obsessed with the idea of like where are you supposed to interact with these systems? Because I think these systems are coming and they're coming to every surface imaginable. And I don't know that anyone has quite figured out exactly the right place for them. So if you find it, you got to let me know. Bestie, I gadget is a smartphone.

2026. She's right again, people. Alison, good to see you. Yeah, good to see it. Thanks for having me. All right, that's it for the show. Thank you to Allison for being here. And thank you, as always, for watching and listening.

Quick thing, before we go, just a small favor to ask. So all three of the Verges podcasts, Verge, Virgin History, and Decoder, have been nominated in the Signal Awards, which is a big awards thing for podcasts. It's very cool, very exciting. We're really glad to be nominated. And I'll be honest, it would be really great to win. So if you could go vote for us, the link is in the show notes. Please vote for Vergecast, vote for version history.

And I suppose if you have to, if you just happen to have time on the side, also go vote for Decoder, I guess. Thank you in advance. We really appreciate it. We're grateful to all of you who listen and watch and vote for this. If you have thoughts, feelings, concerns, if you are mad at me for using Muse and want to talk about it, I would absolutely love to hear from you. Vergecast at the verge.com is the email. 866 version 1.1 is the Vergecast hotline. Send us everything.

I want to know what you're using these things for. I want to know what you feel gross using these things for. I want to know if you just won't use them on principle. This stuff is so new and so different. I want to hear all about it. Get at us. And as a reminder, the best thing you can do to support all of this always is to subscribe to the Verge.

Theverge.com slash subscribe. It gets you all of our podcasts ad-free, including this one. It gets you all of our subscriber-exclusive newsletters. Everything the Verge does. Theverge.com slash subscribe. Thanks in advance. The Vergecast is Verge production and part of the Vox Media Podcast Network. This show is produced by Josh Cahas, Eric Gomez, Brandon Kiefer, and Travis Larchuk.

We will see you tomorrow. Rock and roll.