A billion fake dollars are at stake in our first vibe-coding competition | The Vergecast

You can buy a Google Book. They're on sale. These laptops are the air apparent to the Chromebook, but Chrome OS is out and Android-based Google Book OS is in. The five different Google books on sale now are made by five different companies, and none of them are Google. Starting prices run from $8.99 to $12.99. Dell and HP's Google books run on Qualcomm's ARM-based Snapdragon processors. Acer, Asus, and Lenovo's have x86 Intel chips, but heads up on those. Google already told Android Authority that not every Android app may run smoothly on Intel chips from the beginning. Speaking of Android, Google quietly bumped up the price of the Pixel 10a by $100. The 10a phone will now run you $599 if you buy it MSRP. Of course, Google is not alone here. Last week, Samsung and Huawei bumped up some of their phones' prices. Raspberry Pis are getting more expensive. Not even the 7-year-old Nvidia Shield TV Pro streaming box is safe. And finally, the first new Star Wars racing video game in decades comes out tomorrow, and childhood and current me are both very excited. It's called Star Wars Galactic Racer, and it's getting a lot of love from critics. The Verge's Jay Peters says it's his childhood pod racing dream come true. You can read more at theverge.com. That's 90 seconds on the Verge for Monday, October 5th, 2026.

Hey.

Can I — should I prepare you for this thing that is coming?

Okay, so the show is called Bear Lair. It has no resemblance to any other show that anyone has ever made or seen on television. It's very important that you know that this is a unique new concept that we've created. And basically the idea is you and I in this — in this situation, I suppose, are — we're the bears. And uh, we each have, um, I believe it's a billion dollars in investment capital at our disposal, and we're about to have...

We're going to go. It's a very short episode. Um, a bunch of people are going to come on and they're going to pitch us on the things that they've been building, and we — I think of this as sort of a combination of uh, Shark Tank and The Voice, where you're — we're — we're going to be sort of in on everybody, but you can be in at whatever level you'd like to be, and we're going to bid against each other, and whoever — whoever gets the most investment wins. So, we're gonna get pitches from people. We're going to decide what we like. We're going to be ruthlessly cruel to all of the ideas that we don't like, and then we're going to get into it. Let's bring in Travis, our producer and the — the true hero of Bear Lair. Uh, Travis, this is — this is your baby. Walk us through it. What are we...

Hello, Bears. Uh, as David has described, you each have $1 billion to invest in a number of vibe-coded projects that our lovely, brilliant listeners have taken time out of their days to share with you. They are across the world.

Literally two continents' worth of listeners are represented here today. Uh, and uh, just for — for recordkeeping, a billion dollars is 1,000 million. Please only bid in increments of $1 million.

All right, let's get this show on the road. So, each of — each of our listeners are going to give you a little uh, two-minute elevator pitch, and then we will open things up for Q&A, and then you may bid as much of your fake money as you want. Our first listener joins us from Washington, D.C. This is John.

Bears. I — um, so hi. Um...

Hi. So, um, I have this — I'm going to step back. So, this is — uh, this is a Skylight, which — uh, I don't know if you guys have ever — um, you're a dad, um, David. So, this is like a thing that normally hangs on your wall, and it's supposed to be able to — uh, like, you know, plan your kids' like, you know, appointments, that kind of thing. Um, and it's basically just an Android tablet without a battery, and it plugs into the wall. And uh, my wife bought this, and I did not like it very much. I did not think it worked very well. So, um, through uh, the use of uh, AI, I — um, got it working, and uh, obviously it's not working perfectly well. But if we go here, um, I can show you, um, you know, there's a dashboard. Um, I can actually, uh, control the lights in my home. And, uh, my favorite little bit is I can actually like do a, uh, grocery list app. And, um, once I've done it, it actually comes up as a little, uh, receipt. And I've made this specifically because, uh, I needed like, you know, I'm at the fridge and I — or my wife usually is at the fridge. She needs to, you know, send something to me to be like, 'Can you please pick this up?' Rather than having one running grocery list, I like getting them as tickets. And so now I get these Telegram messages, and they go into TickTick. Um, but what's really cool about it is it's actually like completely adaptable because it's all just a web interface on the back. And so I can actually like go into Claude, as it were, or whatever, and say, 'Hey, I want an app that does this. I want an app that tells me where the metro is. I want an app that shows me, uh, you know, what the PTO is doing right now. I want an app that sets timers.' Um, and you know, plugs into Home Assistant. So, you know, I want an app that, for instance, like I've plugged my car into Home Assistant. So, I want a tab where I can press a button and it turns on the climate control and tells me how long it'll be to go to work. And if I'm running late, then like, you know, can I get a Lime bike or something else instead? I think it's really cool that it's like — it isn't anything. It's just a blank slate, and you know, through just talking to it and spending a lot of money on tokens, um, you can get it to do whatever you want, and that's what I've really enjoyed about it.

Yeah. And it — it actually hangs on your wall. So, it's got, you know, mounting points on the back, and uh, you can actually — so, it — it looks just like a picture, and you can upload whatever photo you want, and it statically just has a picture frame, and you tap it, and then it, you know, goes to your dashboard.

Yeah. I mean, that's literally where I have it. So, I mean, it's — it's actually like, ah, the fridge is here, and this is behind the fridge. And so, you know, when my wife opens it up and she's like, 'There is no — there are no grapes in the house, and what the hell is going on here?' and she can just angrily be like — and it actually says like who's ordering, you know. So my wife can go and be like, you know, 'grapes,' and then type it in, and then I'll get a text being like, 'Where the hell are the grapes?'

Yeah. Yeah. It's just — it's just a really old, crappy Android tablet.

Well, here's the really cool thing, actually, that I love about this, and I feel David — David might really like this. Um, if I use TickTick and my wife uses Trello. If I add something for her, it goes to Trello. If she adds something for me, it goes to TickTick. And as if you're familiar with TickTick, they live in Trello. They live in the cloud. It lives absolutely everywhere. And so I get it on my watch. I get it wherever. And, um, because all this does actually tick an — um, MCP. So it has an MCP server. There's like a TickTick MCP server, and so like, you know, on the back end there's like a cloud server running on a — uh, it's actually running on a broken — uh, a headless MacBook Air, uh, that I removed the screen from. And, um, so it just pushes that into the cloud, and so I can access it from anywhere. But that's what I thought was really cool about it, actually, is this solved a really specific problem, which is Skylight really wants you locked into this app, right? Um, and into this ecosystem that doesn't talk to anything. Whereas this, I like literally wiped it. Like there's no Skylight on this at all anymore. This is just, um, basically just a — a web — a web page that's running on the server. So it also — what's nice is it doesn't matter how slow this particular tablet is. And in fact, I have an accessory for it, which is an old, uh, like Kindle Fire, Amazon Fire 6-inch, like — but it's like literally 10 years old. And I noticed it's got a — it's a nice small screen, really high DPI. And this can run a server that then goes onto that, and by kind of turning it into a kiosk, I can actually tap it and hang it on my wall and control like lights and stuff in the house. So, you know, it's really adaptable.

Okay, nice. I mean, we have a billion, so I was going to go a little higher. Um...

I — I feel like — yeah, I mean, this is exciting to me. I think anything that — uh, anything where the tech is kind of integrated into your home, and I — I don't use like Notion or anything like that, but if I would — if it were on my wall, you know what I mean? So, this is something that I think like everyone in the family can kind of use, especially since it's ported over to your personal device and the type of systems you like. So, yeah, I'll go up to like 100 million.

I should have gone in at less. Oh, well.

Thank you, John.

All right, next up for Hayden is Sarah joining us from Chicago, Illinois.

Hi, David. Hi, Hayden.

Yeah, thank you for having me.

So, um, I'm a parent of a two-year-old, and for me, the hardest part of parenting is like the cognitive load. It's the million things you have to remember. It's the 200 decisions you make every single day. And for me, where it really hit was 9:00 p.m. when I was trying to pack my son's lunch for daycare the next day, trying to remember what my pediatrician told me, what foods my son's been eating that he actually likes, whether or not I've served him something before, and at the end of it, I — I don't actually know what's in my fridge. So, I created this app. It's a combination meal tracker and logger, and it helps me keep track of what's in my pantry as well as what my son has been eating. So, it works like this. I go to the grocery store, and then I come home from the grocery store, and I toggle over to my pantry app, and I tell it what kinds of things I just bought. Today, I just bought snap peas, and I bought him some apples and some other things, right? And I can import it into my pantry, which keeps track of what's available to me, as well as whether or not my son likes it when I decide to serve him a meal. Um, I can go to the planner tab, which has all of his meals for the day. I can either do this ahead of time, if I'm trying to figure out what I'm going to do for the week, or what more often happens for me is after he's eaten a meal, I'm like, 'Oh, yeah, this is what he ate today.' So, I think today at lunch he had something like, um, a — uh, grapes, and then, uh, you know, he had some meatballs, I'm pretty sure, right? And I can add those to his lunch. Uh, after he's done eating, then when I go back to the main tab, it will ask me how his meal was, and I can tell it how much of that meal that he ate that particular day. So today I know at breakfast he ate all of his yogurt, he ate all his apples, and he had some oats. I can also make a note if I'm running low on something or if I run out of something, which is useful for me later. Um, anytime something is low or out, it goes into a shopping list that I can either look at directly through the app or I can copy and paste into another app. And then for me, the most important thing in terms of trying to keep track of like whether I'm giving him a variety of foods, in the planner section there's a weekly tab which lets me look at the entire week and see very quickly what have I been serving him? Have I been giving him a variety of things, or have I given him chicken nuggets every single day the last week?

I think — I think that's a fabulous idea. So, I'll say first of all, there's already a shopping tab which has things broken out by section. So, when you run low or out, I do think the next step would be trying to integrate this with something that can suggest things. And the nice thing is, um, there's a lot of work already been done out there that said if your kid likes this very basic thing, here's how to slowly graduate them to something that's maybe more adult. So you go from chicken nuggets to a slightly different brand of chicken nuggets to shredded chicken to like a chicken breast with a dill sauce or something on it. So I think we could integrate that in to say, 'Hey, if your kid regularly eats or likes this kind of thing, here's the next step up to get them to move towards what we might call a more adult palate.'

Yeah, it is mostly a mobile experience, although I've set it up so that you can, um, push the data to sort of a central place and then sync it across devices. So, actually when I enter my, uh, shopping list for the — or when I come home from the grocery store and I enter my foods, I usually do that on my computer and then sync it to the cloud. That also lets you sync across devices if both parents are responsible for meal prep or — or grocery shopping or anything like that.

Got it. Okay, here's my question.

Yes.

I like — I kind of want this for myself, honestly, because I had Super Burrito, which is my favorite burrito shop, every day for a week last month, and then I was like, I've got to — got to take a bird's-eye view of what I'm eating. So anyway, if I were a parent, absolutely I'd be using this. Here's my question. Do you think it's possible to, yeah, incorporate more of a suggestion mode on — okay, let's say you're looking at the weekly view and you realize, okay, you know, we've got a few food groups covered, but you know, we're really lacking two other ones. Are there — I would love to see like a suggestion mode of, 'Oh, here are two types of things you could be serving,' not even the graduating thing, but just this is what you could be serving, um, you know, that you may not be thinking of, or based on what's in your pantry, here are some other things you could be making that would like, you know, hit those other food groups. Do you think that would be easy to do?

I think it would be relatively simple. I think it's just a matter of connecting the app with an API key to something like Claude or, uh, ChatGPT, and then giving it that — that underlying instruction to say, 'Hey, on this particular window, this is where I want you to generate these kinds of instructions.' I think that would be pretty simple to do.

Okay, I'm going to — I'm — I learned my lesson here. Here, I'm going to start the bidding at 25 million.

I think...

I'm going to go with, uh, 60 million because, yeah, I think that like, one, this is useful for kids, but also like I would use this for myself. So...

80.

These increments are bonkers. You have a billion dollars.

If I knew how many were left, I would be like — it's — it's scary not knowing. Um, okay. I'm going to do honestly like 115. Like I believe in this one, you know? It's — it's — it's also the cognitive load. Like this is — if I had kids, I feel like I'd be using this every meal because it just — I'm keeping track of too many things.

Sarah, thank you. Congratulations on your new-found millions.

Thank you so much.

Bye.

So, just to recap — recap where we're at, David has so far not acquired any projects. Uh, Hayden is down to a mere $785 million, and she has the — uh, John's tablet from which you can request groceries, and also Sarah's, uh, child meal tracker.

All right, coming up next we have Eddie from New Jersey.

Hey, how's it going, David? How's it going, Hayden?

Yeah. So, uh, I don't really — I guess my — I don't have so much as a front-end app as much as like a backend tool for podcasting, media management, that kind of thing. So...

Yeah. So, I — I manage most of the media at my church. Um, and so I have like the whole archive. Um, and every once in a while I get a question that's like, 'Oh, you remember when the pastor talked about this?' And I'm like, 'I honestly — I vaguely remember that, but that may have been like three or four years ago.' So, I started working on this tool last winter where I could basically summarize and pump all this data into — uh, into a Gemini notebook and then have it search for me. And so, the tool is pretty crude, but I can — uh, I can just share my screen here and show you, uh, sort of the workflow that I have. Yeah. So, this is just a — um, a — a command line interface. It's really just a Python script that I had Gemini vibe-code for me. And all you really do is, uh, you put all of the MP3 files for whatever you want to — um, put into this, uh, in a folder with this script, and you run it. Uh, it uses a Gemini API key, so you need to make your own API key, and then it loads a Whisper model and transcribes, um, each MP3 file sort of one at a time. So, it's going through this sermon called 'Understanding Above Knowledge,' and it'll just go through, and it'll, you know, it kind of keeps you updated that it's not frozen as it goes by. Uh, and it takes a little while on my laptop, but on my desktop it's a little bit faster. And if you're running a GPU, it'll be even faster. Um, but you can load up as many as you want. Just let it run overnight. And then what it does once it has the transcript is it'll — it'll take small chunks of it and send it to — uh, I just use one of the free models that, um, Google Cloud offers. And so it chunks it into about 10 different parts to keep the context window manageable, and then it, um, summarizes each of those sections. So it kind of gets rid of a lot of that transcript, um, and repeated sentences and stuff like that. And at the end it gives you sort of a little summary for each one. It just gives you how many times it sent it to the cloud and does some hallucination checks to make sure it's not like totally making stuff up, because that was happening earlier. Like it was worse six months ago when I was doing this, but it seems to be better now. And then what you get out of it is just this — basically this giant markdown file. So this is the sermon database for my church for the year of 2025. And so it's got all roughly 50 sermons in there. And it's just the title, the date, and then the summary. And then I put all of these markdown files into a, uh, Gemini notebook. And this is shared with, um, most of the people at my church. And they can just go in and they can chat with this thing. I hope none of them use it for pastoral advice. But, um, I give like an example prompt here. Like we can just say like, 'I'm looking for a sermon about how to make important life decisions.' And it'll go through and it'll search through all of these four years of sermons that I have in here, and it'll — it'll pull out, um, specific ones that you can then go find on our Podbean or on our Apple Podcast and stuff. And so I — I, you know, for this audience, I think for me what's most relevant is that, um, you know, sometimes Nilay likes to just like drop these like random Easter eggs on, and I've been listening to the show for like four or five years, and sometimes I still have no idea what he's talking about. So I think it would be really nice to just be like, 'Yo, what did — what — what is Nilay talking about?' And he's always like, 'The OGs will know.' And I'm like, 'Now I can pretend to be an OG.' You know...

If it is a — it seems like since it's summarizing each part, it's not like, you know, you can't search transcripts. So, what if it was a quick like Easter egg that someone dropped in? Would that even make it into the summary? As in, would that show up if you were searching for like, 'Oh, David made an inside joke about like shag rugs,' you know, when was that? Do you think that would make it into the summary and be searchable, or probably not because it's not like a high-level takeaway?

Yeah, it — it really depends. So, it probably won't. Um, although it depends on how long they talked about it. So, I've like — I've — I've tried searching things based on like anecdotes because, um, you know, sometimes you tell a story in the middle of a sermon to — to like show a point, and like it will pick up on that stuff even if it's a short story. But yeah, like one-liners, it might not pick up on that.

Yeah.

Yeah, I mean, I — I think like I — I kind of do this manually moving — like going backwards, but I think moving forward it would be nice to either have it automated from our YouTube stream or even like when we upload something to — we use Podbean as our RSS host. Like if Podbean even just like added this, or we — or here, you guys bought it, made it, and sold it to them, and they just added this to, um, to their, uh, website where you could — you could go to your favorite podcast and there would just be a search bar right there and...

Yeah. So I'm — I'm trying to — I — my goal is to get this up and running in NotebookLM and then see if the website team wants to embed it into the website because we have like a web page with our podcast there. Um, and that would be a good spot to have it, but you know, with a big disclaimer to, you know, not ask for advice, but just to use it to look for past recordings. Yeah.

So, I have spent a lot of time looking for a tool that does almost exactly this thing that you just described, uh, which is like — as — as a reporter, I spend a lot of time talking to people about a lot of things. And I'm constantly like, 'Oh, I was just talking to somebody about this thing.' And sometimes that was a week ago, and sometimes it was six years ago. And going through — I have like full text...

Okay. I love that. I mean, I think it's super useful and I would use it, but I think from our conversations lately, David, this is so useful to you that I'm going to let you have it. Like, truly, I was going to start at like 75 million, but you go for it. You need this.

All right. Yeah, that'll work. I love that. Congratulations on the $100 million.

Simon, I hope you've come up with something unbelievably weird that no one would have ever thought of.