Support for the show comes from Jira by Atlassian. Your work lives in scattered places, and it's not only your team searching for context. Your AI agents need it, too. Jira taps into Atlassian's teamwork graph to bring that context together across Jira, Confluence, Slack, GitHub, and more, and makes it easy to share with agents you already use like Claude, Cursor, and GitHub Copilot. Assign them work directly or connect your agents through Atlassian MCP. Learn more at jira.com. That's j-i-r-a.com.
Hello and welcome to The Vergecast, the flagship podcast of everything after the dot com. I'm your friend David Pierce, and today on the show: can you build an app that works inside of a URL? And what does that mean about where the computer is? But first, here's everything else happening on The Verge today.
I'm Travis Larch, and this is 90 Seconds on The Verge for Monday, October 26th, 2026. The iPhone duo is in the wild, and all the internet is talking about is Crackgate. Some customers are posting pictures of inner screen damage, apparently caused by a defective hinge. Hard to say how widespread the issue is, but if I were Apple, I'd hope to have two new product announcements lined up this week to take people's minds off of this. In AI news, looks like Anthropic will beat OpenAI to an IPO. It filed its S-1 prospectus with the SEC today. Did Anthropic use Claude to fill out the form? We don't know. What we do know is that Anthropic took an $80.7 billion net loss earlier this year, and that Anthropic has deals to spend hundreds of billions of dollars with SpaceX, Microsoft, Google, and Amazon over the next decade, which, as you'll recall, is the same time frame some Anthropic researchers say AI could lead to the end of humanity. So spend away. From IPO to IP67, Apple put out a new water-resistant iPad Mini today. It's also got a bigger 8.5-inch OLED display. It's thinner and lighter, and it's $100 more expensive than the last one. But this is the opening act to Wednesday's main event: the rumored announcement of the long-awaited touchscreen MacBook Pro. That's 90 Seconds on The Verge for Monday, October 26th, 2026.
We talk a lot on this show about where the computer is, especially in a world of AI. Is the computer your computer? Is the computer in some Amazon server somewhere? Is it a thing that you self-host? Is it this unknowable set of agents and AI systems that just become a black box you pour all of your data into? What is the right answer? How is any of this supposed to work? And in a really important way, where is my stuff? That's a hard question to answer in the year 2026. Paul Ford is the co-founder of a company called Aboard, working on AI software and all kinds of interesting stuff. He's also a blogger and thinker and writer who has been trying to figure out how technology works for just about as long as anybody. He recently launched a thing called tinylittleoffice.com, and the whole conceit, as you'll hear, of Tiny Little Office is that you can put an entire document, an entire deck, an entire Excel spreadsheet in the URL itself. The URL is enormous and long and contains all of the data that matters. That means it's not on a server. That means Paul can't see it. It's all in the URL. What does that mean about where the computer is? And what does that mean about how we think about the way our data moves around the internet? This made me think a lot of things and feel a lot of feelings about where the computer is going forward. So I figured I'd have Paul on and we'd try to talk through it. Here it is: Paul Ford, welcome back to The Vergecast.
Ah, it is so great to see you, my friend. How you doing? I'm good. Sometimes I just have big feelings about software, and then it's like, well, it's time to call Paul and talk about big feelings about software.
You know, there was a point where somebody asked Joe Pesci, like, 'Do you mind playing the same Italian character over and over again?' He's like, 'I love it. What are you talking about? If that's my brand, I'm going to take it. I like to think about software.' So, what's up? What are your big thoughts?
I want to know about Tiny Little Office. I have a big galaxy brain theory I'm going to put in front of you at the end of this, but at the beginning, you released this little tiny piece of software that is very simple and doesn't do very much, and it's kind of having a moment, and I just want to talk about it. So tell me a little bit about where Tiny Little Office came from.
Okay. Around six years ago, my business partner Rich Zotti, who is still my business partner even though we started a different company — we sold the last one — I'm here at a company called Aboard. We deliver software. I'm not going to market it, but like, we build things. And so one of the things I like to do when I'm running a company that builds things is build things instead of just talk about them, right? And so, and vibe coding, we've talked about it before, it makes it easy. So, you know, a couple months ago, I started to put my — I was like, you know, enough talk, let's start to make stuff. And I've put out a couple little things into the world. And Rich, six years ago, had this wacky idea. It was like, let's make a spreadsheet called Tiny Sheet that you just enter a couple numbers. It was 10 rows, two or three columns. That was it. And it did just a little bit of math. But here was the thing: it was on the web. You went to tinysheet.com and you entered your numbers. But then the URL, the actual link, got the data as you typed it in. It never actually saved anything to a server. It just changed the URL. And so you could cut and paste that into Slack or into anything, and everybody could share this little teeny tiny spreadsheet. So fast forward, that fell off the web, other things happened, and I was like, you know, in the new world of being able to build things and the browsers being faster and better and just sort of the web having come a long way and Claude being so good, etc., etc., I wonder how far you could take that idea. And it turns out you could kind of put your foot on the gas and take it really far and build an entire office suite, including presentation software and including a word processor and a spreadsheet and an outliner and other stuff too, and it could all kind of still be in the URL and be cut and pasted. So I made it and I put it out, and people were like, this is really, really weird.
I mean, it is really weird. Like, tell — okay, okay, just to back up one more second: what does it mean to put all of the information about something into the URL? I think we think about URLs as addresses, right? They are sort of pointing devices; they tell you to go somewhere. I don't think we think about URLs as stores of data in the way that you're describing, but they are in fact that if you want them to be.
Well, there's a couple funny things. So first of all, right, a URL is just data. It's just a string of characters, right? So if you have an infinitely long string of characters, you can store anything. Now it turns out that the web, you know, wasn't designed for an infinitely long string of characters. At the same time, it turns out that the modern web and a lot of social media platforms kind of don't care that much. So you can actually treat like a tweet or a Bluesky post almost like it's a floppy disk. You can just kind of write a whole bunch of data into those things. So in this case, tinylittleoffice.com slash, and then there's a few things that say what kind of application it is, like is it the word processor, is it the spreadsheet, and then there's a whole bunch of gobbledygook. But the gobbledygook is not — it's just sort of compressed, squeezed text. It's not secretly encrypted or anything. It can be — you can actually put a password in, but it's just a bunch of data that is what's in the document. And I did a whole bunch of little tricks to keep the documents very simple. Like there aren't a lot of fonts. You have to really stick to one or two kinds of formatting. It's actually Markdown underneath. Like it's a really compressed version of Markdown with a couple little extra features to make it more like an office suite. So there's a little bit of a parlor trick there, but it's all public. It's all documented, which meant that when you give somebody the link, you're actually giving them the whole document. And let's — you know, one thing you can do with links is turn them into QR codes. QR codes can be quite long. They can be like a couple thousand characters. So you can actually make a QR code that isn't the link but is the document. And that, I think, just the ridiculous puzzle of that, like the way that that's different than what people expected, but it also has kind of a funny payoff because you're like, whoa, I just sent somebody a URL, but that's the whole document. It kind of caught on a little bit. It was fun. It's been really fun there. A lot of people have come to look at it. It's, you know, doing good on Product Hunt. Like it's having a moment, which, you know, I know you as a creator too know, like every now and then you're kind of trundling along and you're like, do I still have it? Can I still reach people? And in this case, it turned out yes. Like hundreds of thousands of people really wanted to look at this thing and understand what made it tick.
Yeah. Well, I think there is something to just the shape of it that is really interesting, right? And you sort of alluded to this in a blog post you wrote about it, which is that we're used to software being like big and packaged and full of lots of things. And when I install something, it's enormous and it does a lot of things, and I have to subscribe to someone's whole idea about how all of this is supposed to work. And every time I do something in it, it is being stored and saved and tracked in some way. The promise of this, at least the way you describe it, is that it's none of those things. It's tiny. It's disposable. Nothing has any idea about the way that it's going. I think if I understand this correctly, the way that browsers and URLs work is that everything after the pound sign means it's not being stored anywhere. It's just being stored in your browser. It's just a whole different way of thinking about what my software is and where it lives. Stored in this — like, I copied and pasted the URL in from one of the decks, I think it's the thing explaining Tiny Little Office privacy, into Google Docs just to see how long it is. It's 3,400 characters, this one URL, and it's nonsense. You can't — it's not like a readable document in this URL, but it contains all of the data that you need to run the app with your data in it. And that is just such a different way of thinking about where my information lives than any other product I think I've seen before. And there's something novel about it.
So there's a few things. There's something you just said that's really important for people to understand, which is you put that pound sign in the URL, and by default the data that goes after the pound sign does not go to the server. So actually, when you go to Tiny Little Office, it loads a bunch of code from Tiny Little Office. Then in your browser, it looks at that data, doesn't send it to the server, and in your browser, it turns it into the document. So your stuff never actually goes anywhere. So if you — the use case everybody has in their head is if you go into incognito mode, so it's not going to save the links that you visit or anything like that, and you put in one of these links, or you write a document with Tiny Little Office and you cut and paste the URL into like Signal, something nice and encrypted, and the person on the other side opens it up in incognito mode in their browser, it doesn't go anywhere. It just literally kind of doesn't exist. You're just sort of passing a secret thing around in an encrypted environment the whole way.
And it's fun to make that a little more accessible to people. It's still pretty complicated to think about, but it's at least a little more instinctively confidential.
To be clear, if I was nefarious — it's one of the reasons this is open source. If I was nefarious, I could, in the JavaScript, in the code that runs in the browser, I could take that data and then I could still send it to some other server somewhere. Like, this isn't like this instant magic security thing, but it is — you can see sort of where all the pieces are. You can understand what's going on, and you know, it's — I did it. My name's on it. So there's a little bit of vouching there, too. But that's real. It's just very fun to make something. There's two things that I'll just say — there's like two things that really mattered to me. One is like, it can be this way. It can be this light. Why don't we make more things like this? And the reason is because they used to be very expensive to make, and you had to figure out — servers needed to be there, and like this kind of play was pricey before AI. It's hard to make stuff like this.
Yeah. It's just not as hard as it used to be. And so you can experiment and fool around and be like, 'Hey, by the way, I made an office suite.'
That's like the nana's thing to say. Like, it has a trip planner, you know? It's got an outliner, but sure, why not? Yeah. I mean, we've been asking the question a lot on this show recently: where does the computer live? And I think part of the reason this fascinates me the way that it does is it is a different idea about where everything lives. And in this case, it's not the computer, but it is the stuff, right? Like, what all the important things about this app live in URLs.
That's right. And in your browser, and they are yours to do with. Now, you can also — it's got a few features, like you can turn on a save mode that saves into the storage on your browser. So, like if you go home and use it, you won't be able to see the stuff that's on your phone on your laptop, right? Like, so there's a lot of little subtle things like that. It also does have a link shortener, but the links are encrypted and the key comes after that pound sign. So your password's actually in the link. You can share it, but the actual stuff that hits the server is completely encrypted. So amazingly, hundreds of thousands of people are messing around with this thing. I have absolutely no idea what is inside. I don't know what they're doing unless they post it on Bluesky. I do see that there's a lot of usage in many different countries, and that's all I know.
I'm just trying to think about how different a way of thinking about data and stuff this actually is, because on the one hand it is by definition more private and more ephemeral because it lives in my browser, it lives on my computer, I can't send it to you if I want to. On the other hand, if I just paste a URL into a Bluesky post, I've now given you everything. And on one hand, that's such a different way of thinking about this stuff, right? Like the traditional thing is even if I sent you a URL, you're not going to have access to my data without a username and password, or if I've explicitly shared something with you. We just have all these structures about where data lives and then how we share it with each other. This turns all of it on its head in a way that is, I think, really useful in some ways, but also just so different that if I want to share something with you, I'm actually sharing it in a more open way in some ways than I have.
Look, it's exactly that provocation. That's what I'm messing with. That's what I wanted. What does it really do when you do something like this? It gets rid of the concept of ownership. You can't own one of these documents the minute that URL gets into somebody else's hand. So one of the examples I like to use is you could print it as a QR code. The QR code now has the entire document inside of it. Somebody else takes a picture. You can't own it anymore. It's their document, right? It's on — and Tiny Little Office loads in, it evaluates the string that's in the QR code, and now it's their spreadsheet. It's their presentation. This is not for your day-to-day office work. It's for pasting stuff into chat. It's for ephemeral stuff. It's for stuff where you want to have just a very light privacy layer because you're dealing with, I don't know, private stuff, or you just don't want stuff saved. You know, the other thing too is just like, you know, one of the wild things about Google is like where it doesn't care — and where it doesn't care is — yeah. No, but like Google Drive, sortability — like when you go into that Google Drive and you're like — I have failed for 10 years. That's what I see when I open Google Drive. Like I'm just — and it's bad, right? And so I'm just like, what was very satisfying was to not leave a mess, to not — to make a little doc and just know that it's gone.
Support for the show comes from Jira by Atlassian. Your work is scattered. A thread here, a DM there, five different docs related to the same project. And now it's not just your team hunting for context. It's your AI agents, too. Jira taps into Atlassian's teamwork graph to bring that context together across Jira, Confluence, Slack, GitHub, and more, and makes it easy to share with the agents you already use like Claude, Cursor, and GitHub Copilot. Assign them work directly or connect your agents through Atlassian MCP. The result: less token waste, more accuracy, and more time moving work forward. 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 j-i-r-a.com.
I mean, this goes back to what you were saying about kind of the advent of AI and vibe coding. It does feel like the reason this kind of software doesn't really exist is that the gap between how hard it is to make and how much sort of revenue potential there is in it has always been quite large, right? You spending a lot of resources to build ultimately pretty small software is a bad trade in a lot of ways, unless it's just a thing you do for the joy of doing it. In this case, you've brought the effort way down. So the idea of using small effort to build small software suddenly makes all of this stuff possible, right? Like, this is the kind of thing that I'm sure you're not going to spend two years hand-coding just to see what happens.
I spent 10 days, and most of those were sessions where I was on the train going home. And I think like the last couple days I really did kind of do a lot of loops to get — but no, but like maybe 20, 30 hours total, right? So, like, let's put that into context. It's not me saying what a good boy I am.
The reality is that you can just — if you think hard about the architecture of something and you really structure it. So I did a lot of like little tricks to test it, to make the way that the commands work as elegant as I could, so it could automatically evaluate itself. And then there's another thing that's really important: is software at some level, like what's inside of a web browser today, like Chrome or Safari, is so good, it's so mature, but it's all architected kind of for big companies. Like most people, including me, have very little direct experience with the encryption tools inside of a browser. Sure. Right. They're absolutely first class. And so I was like, 'All right, I'm not going to encrypt anything. I'm an idiot.' But Safari and Chrome and WebKit, though — they can't. And I trust them. I give them my credit card all day, and so does everybody else. I'm like, 'All right.' So I'm talking to it because that's what you do. And I'm like, 'Okay, let's make sure this is safe.' And it's like, 'Well, if you want to be really safe, we got to do the AES-256 post-quantum encryption.' And I'm like, 'We do. We have to do that.' And it's like, 'Okay, I did it.' And I'm like, 'Okay.' Like, I'm learning as I go and I'm getting caught up. But what I also know as a programmer is that if I'm making everything into one big set of characters, I'm going to encrypt only that set of characters. So I kind of know where my encryption and my security is. And so suddenly, because the browsers are so good and mature, and because the vibe coding tools are like — I can kind of narrow down the risks in a platform like this, and I can just kind of say with high confidence that when you put a password on one of these things and you encrypt it and save it on the server or pass it around, it's pretty encrypted because it comes out of the browser's own, right? And you can go look at that code. It's like one JavaScript file. So anybody who wants to beat that up, I would really welcome it. Like, 'God, if you could find that bug, I would love it.' And so that's a very different world because it turns out that everything is really, really good, but has been really hard to get to because it's for Google. Like, Chrome is for Google to do video streaming and encryption and transactions. And now it kind of can be for everybody. And I want to throw this at you because I just want to — in your head in particular, it's as good as any other platform. It's got 3D, it's got, you know, networking and audio and so on.
There is very little reason for mobile to be the way it is in terms of the app stores.
In this era when you can target even a mobile web browser like —
Sure, I mean, you know, the reason is money, Paul. I don't know if you've heard of money. Google and Apple have a lot of it, and they would really like to keep adding more of it.
It has something to do with capitalism, but I've never been able to like put it all together.
