AI didn't replace me - it made the case for me
Why AI has made senior craft, judgement, and useful product thinking more valuable, not less.
The last two years have produced more websites than the previous ten combined. Most of them look the same. Same hero with a gradient blob. Same three-column card layout. Same “Book a call” button in the top right. Same copy that could belong to any company in any industry in any country.
That’s what happens when you hand a brief to a machine that has consumed the entire internet and been asked to produce something professional-looking. It produces the average. Competently. Instantly. For almost nothing.
I don’t say that as a complaint. I say it as an observation that has directly changed the value of what I do.
There’s a version of this story where AI collapses the market for independent studios like mine. Founders build their own tools. Agencies scale headcount with AI agents instead of junior designers. The work gets commoditised and the price goes to zero.
That version exists. It’s real for a certain kind of studio - the one doing repeatable, templated, brief-in-brief-out work. That market has compressed badly, and I think it probably should have.
The version I’m living is different.
The clients coming to Dream Creative in 2026 are coming specifically because they’ve already tried the fast route. They’ve got a Webflow site that looks fine but doesn’t convert. A prototype their developer shipped that technically works but feels wrong. An MVP that’s live but doesn’t communicate what the product actually does. They’ve seen what fast looks like. They want something that actually functions.
That’s the distinction that matters: fast versus useful.
I’ve been building things for ten-plus years across three disciplines - web development, 3D rendering, photography. In that time, the tools have changed completely. The problems haven’t.
The problem is almost never “this project needs better code” or “this project needs a better visual.” The problem is usually that the people who built it got too close to it, too fast. They stopped asking whether it was useful and started asking whether it was done.
AI is very good at “done.” It is not yet good at the harder question, which is: does this actually work for the person who’s going to use it?
That question requires sitting with the messy prototype for long enough to understand what it’s trying to do. It requires reading the data, looking at where users drop off, understanding the council dataset that powers the rubbish collection calendar well enough to know why the Wednesday result is wrong for certain postcodes. It requires having an opinion about what the right fix is - not just generating five options and presenting them neutrally.
An opinion is the thing AI can’t quite do. Not a real one. Not one that costs you something if you’re wrong.
The studios I’ve seen struggle most in the last two years aren’t the large agencies - they’ve absorbed AI tools into their workflow and kept billing. The ones struggling are the mid-tier generalists. The studios that were competing on breadth, quick delivery, and a reasonable rate. That middle got cut out. They couldn’t compete on speed with AI, and they couldn’t compete on craft or depth with the senior independents.
Dream Creative was never in that middle. It was always on the side of depth. One project at a time, full attention, no junior work dressed up as senior. The deliberate smallness that once required explaining - “but it’s just one person” - is now a feature that clients understand without prompting.
They’ve seen what “one person” produces when that person is a mid-level generalist. They’ve also seen what “one person” produces when that person has been doing this for a decade and cares about the outcome more than the invoice. The difference is not subtle.
Here’s what’s actually changed in my workflow.
I use AI tools every day. For code completion, for copy drafts, for quick research, for processing data. I’m probably delivering better work faster than I was three years ago - not because I’ve scaled, but because the tedious parts take less time and the thinking parts get more of it.
What I’m not doing is using AI to replace the part that matters: looking at the actual project, forming an actual view, building something that works for the actual person who’s going to use it.
That part is still manual. It’s still slow by algorithmic standards. And it’s exactly the part clients are paying for.
The best summary I’ve found for what’s happened: AI raised the floor and lowered the ceiling on generic work. Everything average got faster and cheaper. Everything that required real craft, real attention, and a real opinion got relatively more valuable.
If you’re building something useful - a public tool, a product that solves a specific problem, a prototype that needs to become a proper version - the commodity option was probably never right for you anyway.
Send me the messy prototype. I’ll look at the real thing.