GoKawiilGet Your Reps
product, design, and engineering lessons from building Sail & Muddy
In 2022 I was looking for a very specific thing: a small team, less than ten people, before product market fit, working on something I actually cared about. I started coding early on and have always loved building and computers. I was interested in startups and in the future of personal computing. Two friends from college, Ron and Jimmy, reached out. They'd been working on this idea of a new browser, forked Chromium, figured out how to build on top of it, and raised a 5.5 million dollar seed from General Catalyst, Naval, Lachy Groom, YC, and others. By the time I joined as a founding engineer one of the hard parts was done. You could build on top of Chromium, access tabs, the history API, and all the UI could be built with web technologies. But the other hard part was still ahead of us: what to build, how it should work, and whether anyone wanted it.
I've always loved the philosophy of the open web, and I'm deeply grateful for it. I love frontend engineering and I'm very product minded. Getting to build on top of Chromium, millions of dollars in R&D forked from , felt both humbling and like an experience that would really stretch me. We were aiming to build a window worthy of the work you do and the work you do with other people. That evolved into collaborative software: realtime multiplayer, infinite canvases, rich text editors, chat, all packaged into the browser alongside web content. Framed as a "multiplayer browser" or "team browser," it was an attempt to channel venture dollars into a new kind of personal computing company. Sail, and later Muddy, were the products that came out of that vision.
Lots of people want to build new computing paradigms, new interface ideas. We gave it a real shot. Past flashy demos and hypotheticals. We studied both the history of personal computing and your favorite productivity software (Slack, Notion, Linear, you name it), talked to users, built new interfaces, shipped them, and watched most of them not work. Building and maintaining a browser product was unusually hard, and even though it did not work out as a company, the team proved a lot and learned a lot. Building a startup forces you to see past the grand vision. We failed to make something that a lot of people wanted and could grow. But I came out of it a fundamentally better product thinker, engineer, designer, builder, and hacker than when I went in. I got my reps in. Here's what I took away.
What We Built
When I joined, Sail was a very much in progress infinite canvas app built on top of our own . You could put websites in, add some text cards, see people's cursors, but it was still taking shape. Positioning wasn't locked in yet, and there were features to build out and bugs to fix. We iterated on it and drew inspiration from , Kinopio, and a lot of other infinite canvas apps I can't even remember all of, riding the energy of the canvas moment when Miro and FigJam were surging. A fun positioning we talked about at some point was "Spatial Notion." Sail was the most fun version and what got me to join. Using it felt magical. Browsing the web and feeling like it's the web, not a streamed video feed like you would in a Zoom screenshare, seeing other people's cursors and imagining how everyone could actually be doing work in the same canvas side by side, having these random encounters like you often do in Figma when you browse the same design file. It felt super cool. But cool demos, cool visions, and cool feelings aren't always enough. Though of course, sometimes they can be.
Sail never got a broad public launch. We tested it with different types of users looking to make themselves more productive, but there's a chance we failed to reach a specific type of user that would have loved Sail. I really wish we had launched more . No one cares about your launch. Taking too long to launch builds up pressure. You start protecting a reputation you haven't earned yet. And if you take too long, you give yourself fewer chances to relaunch. said it best: if you launch and no one notices, you can just launch again.
After Sail we went through what we internally nicknamed the "multiverse project," a version that supported different kinds of boards: infinite canvas, structured canvas (Nototo-style), and chat. Because it was all built on the same sync engine, trying different form factors was relatively easy. We even got to the point where you could reference usable instances of one board type inside another. It was cool. It also got really complex. Chat is what survived, and it became the foundation of Muddy.
Chat is . People understand it immediately, it's legible, it caters to a lower common denominator more easily than a canvas, and it transfers to mobile. We had a React Native app on TestFlight, built with help from contractors. The scope and table stakes for productivity software keep going up, people expect a mobile companion app.
... continue reading