Writing

Writing Four essays on shipping, design and code.

Tags 10 tags across four essays.

The portfolio I actually finished

The portfolio I actually finished Every designer I know has a half-built portfolio site.

8 May 2026 · 7 min readThe portfolio I actually finishedMost portfolios live in a graveyard called Drafts. I had three of them in there. A 2022 attempt that died at the hero section. A 2024 attempt with a completed home page and zero project pages. And whatever this is now, except this one is at a real domain, with real projects on it, and you're reading the post that proves it.The reason the first two failed was the same reason every designer's portfolio fails: I kept treating the site as a moving target. Every time I sat down to ship a page I'd find six things I wanted to redo first. New font pairing. Tighter spacing scale. A "better" colour palette. The work expanded to fill the available perfectionism, and the perfectionism was infinite.So this time I made three rules and stuck to them. Rule one: ship one tile a day, even if it's ugly. A bad tile on the canvas beats a perfect tile in my head. The yellow "I'm William" hero went up first, looking nothing like the eventual version, and that one ugly tile was the difference between a real codebase and a Figma file.Rule two: every section gets exactly one idiom. No card on this site borrows another card's treatment. The home page is sixteen tiles and sixteen distinct visual ideas. Boarding-pass academic tickets, sticker company badges, a heatmap ledger, a polaroid carousel, a magazine pull-quote. The constraint is brutal: invent or steal. The reward is that no two cards bore me, so no two cards bore the visitor.Rule three: deadlines beat reviewers. I gave myself a public commit window. The repo went open the moment the home page rendered. Pushing to a private branch and "polishing it first" is the same trap, just with different paint. Once it was public I had to keep going, because if you abandon a public repo you have to explain it.Perfectionism is the polite name for not shipping. The site you're reading exists because I stopped editing the README and started merging. The version you see today is not the version I imagined. The version I imagined still doesn't exist. The one that does is fine, and being fine on the internet beats being perfect in a Figma file.

Framer, as a production tool

Framer, as a production tool What I've learned shipping Studio's product pages on Framer for a year and a half.

12 Apr 2026 · 6 min readFramer, as a production toolFor most of my career Framer was a prototyping tool. You built clickable mockups, you exported them as a sharable link, you got feedback, you went back to React. Framer Sites changed that: the things you make in Framer can be the thing the customer actually loads. That sounds small until you're shipping pages a couple of times a week.The first thing you learn is that "design directly in canvas" is not the same as "design in Figma, then implement in code". It's a third workflow with its own muscles. You have to think in components from the start, because reuse on a Framer page is not a refactor, it's a primitive. Build the component once and your three product pages have the same hero rhythm without you trying.The second thing is motion. Framer is generous about animation in a way React is not. You can scroll-link a transform, you can stagger children with one number, you can easily wire a hover state that shifts three properties at once. The temptation is to use everything. The discipline is to pick one or two motion ideas per page and let them carry the whole thing.The limits show up where you'd expect: deeply dynamic content, complex auth flows, heavy data tables. Framer is not the place for those. It is, however, an excellent place for the marketing surface that wraps them. We separate concerns: Framer for landing pages and product marketing, our React app for the actual product, and the seam between them is just a link.A few smaller lessons I keep relearning. (1) Variables save weeks. Set up your colour, type, and spacing tokens before you make a single component. (2) Breakpoints are state. The mobile layout is not a smaller desktop, it's a different design with the same words. (3) Scroll position is content. Treat the scroll like a timeline, not like a long page.If you're a designer who can read code, Framer pulls you into shipping. If you're a developer who can think in design, Framer pulls you into thinking about what the page is for. Both are useful pulls. The work I make there is more thoughtful than the work I make where the canvas and the codebase are different rooms.

The cats built the studio

The cats built the studio Mochi naps on the keyboard, Kopi judges every PR, Susu owns the radiator.

22 Mar 2026 · 4 min readThe cats built the studioThe studio has five employees and only one of them is human. Mochi is the office manager: adopted in July and already territorial about the chair. Kopi is the early code reviewer; he sits on the second monitor and stares at every diff like he's catching the typo I just missed. Susu owns the radiator and considers her job done. Bumi turns up for the morning standup and disappears for the rest of the day.Working with cats has done two things to my pace. The first is the most obvious: the day has more interruptions. A cat who wants attention is not negotiable. The keyboard becomes a bed. The mouse becomes a toy. The Slack notification becomes the sound of someone judging you for taking your eye off the bowl. You stop more often than you'd plan.The second is less obvious and more useful: the day has more pauses. A cat sleeping on your forearm is the most honest deadline you will ever meet: you will not finish that ticket, you will sit there, you will look at the wall, and your brain will quietly defrag. Most of my best decisions about a UI started during one of these forced pauses, not during the active typing that bookended it.I used to think interruptions were the enemy of deep work. The cats taught me to separate interruption from decision-making. The interruption is the cat. The decision-making is what happens while the cat is on you. I stopped treating cat-time as a tax on output and started treating it as the room where the work actually thinks.The studio is small. The cats outnumber the humans. The bug list still gets shorter every week, and somehow it feels less stressful than the years I worked alone with no fur involved. There is probably a productivity book in this. There is definitely four bowls to fill at five.

Draw before you code

Draw before you code Why I still start every new feature on paper.

8 Feb 2026 · 4 min readDraw before you codeMy favourite part of starting a project is the part where I don't have a tool open yet. There is a notebook. There is a fineliner. There is, sometimes, a coffee. The rest of the day will be screens, but the first hour is paper. I don't think this is precious. I think it's load-bearing.Drawing forces commitment without polish. When you're sketching a layout you can't accidentally fall down a font picker, or get distracted by whether a corner radius should be 8 or 12. The only knobs you have are: where things go, how big they are, what's near what. That's the design problem. Everything else is finishing work.The other thing paper does is accelerate divergence. I can sketch six layouts in the time it takes Figma to render an autoLayout reflow. Most of those layouts will be bad. That's the point. The bad ones are how you find the good one: you only know it's good because you can compare. If your first layout is the only layout, it's just an opinion.There's a counter-argument that AI tools (Cursor, v0, the latest Vercel thing) collapse this loop. You describe a layout, you get four versions, you pick the good one. I use those tools. They are good for the part of the work where you already know what you're trying to make. They're worse for the part where you don't yet: they will gladly generate four polished versions of the wrong idea.So I keep the notebook. The first hour is paper, the next nine are Figma and Framer and a code editor. The work that comes out feels like one decision instead of a hundred small ones, because I made the big decision when there was nothing else to look at.

Sources

Sources What the essays link out to.

Keep going

Keep going The rest of the site, one click away.

AboutSince 2018Who builds them, and where.Go to About