Everyone would fall asleep and the house would go quiet. I’d come back to my little workspace, dim the lights, put on my favourite music. Two shots of espresso on ice. Dark chocolate within reach. Then I’d open the file and pick up wherever I’d left off the night before. Most nights I got to bed around three.
That was the whole production schedule. An hour here, an hour there, the way the name suggests. A newborn, a new rhythm at home, and a full plate at work meant the long focused stretches were never coming. If a new site was getting built, it was getting built in the leftovers.
I liked my old site. I built it in 2021 and it did one job well. It was close to an artist statement. Minimal, plain, here is who I am. But somewhere in the last five years it stopped telling the truth. My work had always run on two tracks, design and stewardship, and the second had only grown, but the site showed just one of them. It held what I’d made. What I had to say didn’t have a home yet.
I tell mentees all the time that keeping your personal brand current is one of the most important things a designer carries. Mine was five years behind who I’d become. I heard myself every time I said it.
And there was one more thing. I’d gotten good enough at vibe coding that I started to believe I could build the whole thing myself. Every decision, every line, mine. Scale it however I want, without waiting on anyone. That thought was most of the fuel.
My first approach was to jump straight into Claude Code and just start exploring. It felt fast. Night after night of trying ideas in code, and what I had at the end was a polluted codebase and a direction I didn’t recognize. The tool kept moving, and I was the one following.
So I started over. A rough design system first, in Figma. Colours, typography, spacing, all set as variables. Still evolving, but drawn. Then Figma Bridge into Claude Code, building exactly what I’d designed. You know what you’re getting. Unless every idea is already solid in your head, draw the thing first.
Knowing how to code helped more than any prompt trick. I could tell it exactly what to build and how, and it did.
Then the season turned heavy. Work was going through changes that drained me more than I wanted to admit. The kids needed everything kids need. I was tired in a way sleep wasn’t fixing, and some nights a portfolio redesign felt like a luxury I had no business taking.
I showed up anyway. Every night. Some nights the progress was real. Most nights it was design, undo, redo, pull an old idea back, undo, try something new.
Then one night it clicked. The type, the spacing, the colour, the patterns stopped fighting each other and started reading as one flexible system. I sat there in the middle of the night smiling at the screen. A big smile. I couldn’t remember the last one. It felt like getting my designer-self back.
One rule sat over all of it. No matter where you land on the site, it should read as one designer. So almost every component is reusable, every detail considered against the system it lives in.
Here is the part I keep thinking about. These tools will hand you a competent site all night long. Clean, current, fine. That was the trap, and I fell into it more than once. Every time I took a suggestion without asking why, the site drifted a little further from mine and a little closer to everyone’s. What pulled it back was the years. Knowing what good looks like, why it’s good, which good is mine. The tool was a real collaborator, but the designer part was still my job.
The clearest line I drew was in the work pages. Every asset in them crafted by hand. No AI in any of it. That’s where I went fully expressive, and the strict system around them is what held it together.
The strange part is what the schedule did to the work. I assumed building in leftovers would be the project’s biggest setback. But an hour a night is too slow to rush anything. It kept me sitting with the harder question underneath the site, who I actually am as a designer and an educator, long enough to answer it honestly instead of shipping the easy version.
By the end I’d been staring at it so long I was starting to lose my mind a little. I’m a designer at heart. I could iterate on this every night for another year and call it progress. So I proofread everything three, four times, made sure every line read the way I meant it, and soft launched. Then I made myself not look at it for three days.
When I came back I spotted some visual bugs and the sentences that didn’t sit right, the ones I couldn’t see before. I fixed them. A few days later I announced the new 5am.co.
The site finally says what has been true for years, that the two tracks were always one work. It says who I am right now. Stopping was the last design decision, and the hardest one. It’s made. For now.