New website, new beginnings

Since my very first portfolio, I used Webflow, and later Framer.

Those tools felt like they were born for designers. An intuitive canvas I already knew from Figma, a single button to go live, and a price point that felt a lot more reasonable than contracting a developer to build me a site I wouldn't be able to touch myself.

Over the years I shipped more than ten different iterations of my portfolio, all of them in no-code. For the most part, everything was fine.

What changed

In 2026 I stopped being "just" a designer and started working as a design engineer, shipping with Claude Code and Codex.

That's when the absurdity of no-code became hard to ignore.

On client projects, I'd open Figma to draft something and then produce the final design in code. The workflow made sense. I kept agency where it mattered, and handed the implementation to the agents.

But on my own site, every change meant going back to doing everything by hand. Things that were trivial during client work turned into an afternoon in Framer. Maintaining a portfolio is annoying enough on its own. I want to spend that time on client work, not on managing my site.

The site you're reading this on is my first coded portfolio.

How it's built

It's a Next.js site on Vercel. No CMS, no database, nothing to log into.

Everything lives in a handful of folders:

work        one entry per project
writing     one file per article
screenshots one folder per case study
components  the buttons, text and layouts everything is made of

That's the whole thing. If I want to change something, I change a file.

The design system does the boring part

I ported the Figma library into the codebase once: colors, type, spacing, corners, shadows. Every component pulls from it, so there's exactly one way to build a button, a heading, or a card, and light and dark mode come for free.

The point isn't tidiness. It's that Claude Code already knows how this site is supposed to look, so a new page takes minutes instead of an afternoon of dragging things around a canvas.

The case study workflow

Here's where the difference actually shows up.

Say I just wrapped a website project. In Framer, that meant visiting the site, taking screenshots, cropping and exporting them, dragging them into a CMS collection, and writing the copy field by field.

Now the page builds itself from a folder. I drop images or videos in, named in the order I want them:

work/pinetree/
  00-hero.mp4
  01-dashboard.png
  02-episode-page.png

They show up on the case study in that order, at their own proportions. No slots to fill in, no layout to rebuild.

The copy sits next to it, as a few lines of text:

Pinetree
2025, website and product design

Pinetree is a producer agent for weekly shows...
I worked directly with Giuliano, their founder...

So the job becomes: ask Claude to visit the site, capture the screens, and draft the case study. I review whether it makes sense and step in where it doesn't. Same shape of work I do every day, on the same tools.

The best part is that I can do it for six projects at once, in the background, while I work on something else.

On no-code

I think no-code's golden years are behind it. I still have a lot of affection for what those tools let me do, and I wouldn't have had a portfolio at all without them. Today they just don't make sense for how I work.

The obvious objection is that no-code tools have AI now. Fair. But if the AI is doing the building either way, what exactly are you paying for? A visual canvas and a web-design-shaped wrapper around a codebase you don't get to keep.

This is also what pushed my brother and me to start Port. We take companies off an expensive, agency-managed no-code site and move them into a codebase they own, one their own team can edit without waiting in anyone's queue.

Eleven portfolios in, this is the first one I'm not dreading updating.