Automated Image Sizing – Built in
Using the right CMS will save you resources by automating and serving optimized content for your website. Now that's performant!
Good side effects
One of the more interesting side effects of moving away from traditional, canvas based workflows and toward a production-code as the single source of truth, is that you start rethinking everything.
Even something as mundane (and surprisingly complex) as image optimization.
The problem I thought I was solving
I've always been a big fan of WebP.
My original plan was straightforward: generate multiple image sizes and formats manually, then serve the right asset to the right browser at the right time.
Sensible enough. Then I took a closer look at the stack I’ve assembled and combined.
The infrastructure was already smarter than I was
Because I'm using Sanity as both my CMS and image CDN, it turns out I can upload a single high-quality source image—currently testing around 2,000px wide, exported as WebP at roughly 90% quality—and let the platform do the heavy lifting.
Sanity dynamically resizes images based on the request, serves AVIF to browsers that support it, falls back to WebP when needed, and only delivers the exact asset required for that specific device and viewport.
I used to spend hours on previous projects, that modern infrastructure had already solved better than I could have.
And honestly? I love that ♥️
The single source of truth strikes again
Combined with Astro on the front end—delivering highly optimized, static-first experiences—the result is a stack that feels incredibly fast, incredibly flexible, and surprisingly elegant.
Upload the right source asset once. Let the system do the rest.
The more I work this way, the more I'm convinced that single source of truth systems aren't just a simple architectural decision.
They're a cognitive architecture decision. This is infrastructure.
The less time I spend managing duplicate assets, duplicate decisions, and duplicate sources of context, the more time I spend thinking about the thing that actually matters–the content:
What am I trying to build?
AI changed how I think about systems
One of the unexpected side effects of working deeply with AI is that I've started thinking less like a designer and more like a systems architect.
Or maybe more accurately: A spatial thinker.
Instead of asking, "What artifact do I need to create?"
I'm increasingly asking:
"What environment do I need to create so that the right artifacts emerge naturally?"
That shift changes everything. The image isn't the deliverable. The workflow is.
The website isn't the deliverable. The system that generates and maintains the website is.
Why I'm moving beyond website builders
This is one of the many reasons I'm increasingly moving away from all-in-one website builders, even the really excellent ones.
I'll absolutely continue using platforms like Webflow and Framer when they're the right tool for the job. They're fantastic products.
But for projects where performance, control, extensibility, and long-term ownership matter most, I'm finding myself drawn more and more toward custom-built systems, more specifically the stack I’ve assembled at Funda Run.
Not because I want to write more code. Because I want more control over the relationships between things.
The future of design
The deeper I get into this world, the more convinced I become that the future of design isn't choosing between design, engineering or content.
It's becoming fluent at all of it. And perhaps more importantly:
It's learning how to build systems that expose intent, preserve understanding, and record the best decisions.