What it actually means to be AI-native.

There's a version of "AI-native" that means: we use ChatGPT for first drafts. There's another version that means: we rebuilt the pipeline around what the model can actually do well. The first version is a marketing claim. The second version is a different studio.

Bolt-on vs. native

A bolt-on studio keeps the relay structure — discovery, design, copy, dev — and adds AI as a tool inside each phase. The model writes a first draft of the copy. The designer uses an AI image tool for moodboards. The developer pastes code into Cursor. The phases stay the same. The handoffs stay the same. The calendar stays the same.

An AI-native studio rebuilds the pipeline against the question: what does the model do well, and what should the human still do? The answer changes the shape of the cycle. Design and copy run in parallel because the model can hold both in context simultaneously. SEO structure and CMS scaffolding generate from the page list because the model has no problem with the boilerplate. The human work concentrates on the decisions that don't generalize — shape, tone, the final read.

What worked

Three things worked immediately. Drafting copy from a tight brand brief — the model is much better at writing in a voice you've given it than at finding a voice. Generating SEO and metadata in parallel with the page build — schema and sitemap are structural, and the model handles structure well. Scaffolding CMS content models from the production layout — a tedious task that used to consume a day or two.

What didn't

Two things didn't work and we stopped trying. Image generation for hero shots — the model is good enough now to be tempting and bad enough still that a human eye catches the artifacts on every project. We use photography or illustration sourced traditionally. And open-ended brand discovery — the model is too willing to converge on the genre average. The brand direction conversation stays human-led; the model joins once direction is set.

The studio is still the studio

Being AI-native doesn't mean fewer decisions get made. It means decisions concentrate in different places. The model does more drafting; the studio does more editing. The model does more structure; the studio does more shape. The work that ends up on the client's site is studio craft, run through a different pipeline. The pipeline is the differentiator. The craft is still the craft.