There’s something about spring that makes you look at old habits and wonder why you were ever doing it that way.

Last piece I wrote here, I was running four Claude Code windows simultaneously and hadn’t built a single thing that worked. I said I was in first grade.

I’ve since moved up a year or two. One of the things that changed most visibly was design.

A Problem Everyone Recognizes

Design has always been the bottleneck nobody wanted to address. Campaigns wait for assets. Launches wait for visuals. Social calendars stall. Everyone knows it.

What changed is that the queue is now optional. Not by replacing designers. By teaching machines to design, and giving everyone in the company access to brand-aligned output without routing every request through a human. Slides, social, video, emails. All of it.

The Toolkit

A few tools made this real. They’re worth naming, but none of them is the story.

Google AI Studio and Stitch are the prototyping layer — a direct interface to Google’s models via API, where you test and refine prompts and design directions before wiring anything into production.

Gemini 3.1 Flash Image handles image generation at the API level. Fast, capable, built to run at volume inside automated workflows.

MCP-connected design and website template services are where generation becomes production. An agent receives a brief, generates design output, and pushes it into a live template. No human in the middle step.

Claude’s frontend design skill produces clean UI, layouts, and visual structures from brand context and constraints. Fast first drafts that are often closer to final than you’d expect.

These tools matter. But what makes them useful is what you feed them.

The Real Work: Teaching Machines to Design

The most valuable thing we built wasn’t a tool. It was a document.

A design skill file — a structured set of instructions that tells any executor, human or machine, exactly how to produce on-brand visual output. Colours, typography, spatial rules, compositional logic, tone of imagery, what to avoid and why.

Not a mood board. Not a brand guidelines PDF nobody opens. A skill. Something executable.

The goal was simple: someone on the team provides a task, links the skill, and the output comes back on-brand. No design knowledge required. No queue.

Most brand documentation is written for humans. Qualitative language, things that “feel like” something. Useful for inspiration. Useless for automation. Machines need specific, hierarchical, unambiguous rules. What’s interesting is that when you write brand rules that precisely, humans find them easier to follow too. Specificity turns out to be a gift for everyone.

The skill file became the single source of truth that both our team and our agents pull from. Brief goes in, skill goes in, consistent output comes out. Whether a person is driving or a tool is.

That’s the unlock. Not the generation. The consistency at scale.

What I Now Know That I Didn’t

• Brand-aligned output used to require headcount. Now it requires a skill file. Every team member becomes a creative director, not a producer.
• Write for machines, get better human output for free. Precision and hierarchy in brand rules makes them executable for everyone.
• A skill file is infrastructure, not a creative deliverable. Version it. Maintain it. It compounds in value every time someone — or something — uses it.
• Format coverage matters from day one. The skill needs to speak to every surface: slides, social, video, email, web.
• MCP connections are where generation becomes workflow. Output lands in a live template, in the right format, without a human touching it.
• The hard part is technical. Treat it like an engineering problem from the start.

The Bottom Line

Every spring you get the same invitation. Clear out what isn’t working and make room for something better.

The design bottleneck was never inevitable. It was structural. And now the structure has changed.

When machines can design from a skill file, the question is no longer “who has bandwidth.” It’s “what are we building next.”

About the Author: Ziv Meiri is a global technology, marketing, brand and creative executive, advising leadership teams on growth marketing strategy, brand positioning, and creative systems.