Design tokens are machine-readable. Your brand guidelines aren't.

We paid a lot of money for a lifestyle brand

We sold professional services automation software. What came back from the rebrand was a lifestyle brand.

It was a big agency and a big invoice, and what came back was beautiful in isolation and completely wrong for what we actually did and who we actually sold to. Then the brand team was cut, and the rest of us picked up the pieces and reworked it for new leadership and a new board.

That left 80 pages of brand guideline nobody used, and we wondered why.

Showing people what good looks like is fine. But unless they know what it takes to build like that, you’re asking for inconsistency. Pages of examples, no instructions.

Hand that to a contractor or an offshore team and every deliverable comes back feeling like a different brand. You’re relying on a designer to study the examples, work backwards to the rules nobody wrote down, and carry the vibe on their own. That also assumes the examples were the right ones. Brand guidelines are full of billboard mockups and tote bags. We were making one-pagers, ebooks and presentations.

The brand itself needs about 10 pages. Split the rest into instructional content. Disclose what people actually need, not what makes the document look impressive. Don’t write a brand guideline to stroke your ego, write one to help people make decent work. Cut the gatekeeping and let the brand serve the people using it, because without them we don’t exist.

Then we paid a different agency to build the design system

This one was for a replatform, and they came in through an existing relationship on the marketing side rather than anything resembling a brand process. They billed a lot and delivered very little.

The colour options came from their marketing consultant picking things they liked. No strategy underneath any of it. Nobody had asked whether these were our competitors’ colours, whether they differentiated us, or what they were meant to make a prospect feel. A colour scale? One weight per colour. Semantic roles? They didn’t know what semantics were. Somebody was fairly obviously asking GPT for suggestions and firing them over hoping we’d agree.

Then the wireframes arrived, and these were the hi-fidelity ones. They looked like a website from 20 years ago. Blocky, no style, every component slightly different on every page even when it was structurally the same component, everything spaced at random.

We had a come to Jesus moment and I rebuilt it. Spacing, components, tokens, a colour system chosen to sit near Salesforce and away from everyone we competed with, and typography that was web safe and ours. After that I did the work and directed the agency we were paying to do the work.

The half that was already solved

The design system side had worked this out years ago. Tokens, scales, components, semantic roles. Written so a machine could hold it and a person could pick it up without a workshop. Nobody had done the same for brand, because brand’s consumer was always a person, and people fill in the gaps.

Then we started handing brand to machines, and the gaps stopped getting filled.

You can spot the ungoverned version instantly

Everything looks the same. It looks like every vibe-coded one-shot homepage.

Container city. Boxes inside boxes inside boxes, every section wrapped in a bordered rounded rectangle. Low contrast text, bad spacing, no hierarchy. One accent colour, light or dark mode, no colour ramps.

Junior design with no taste.

That isn’t a model problem. That’s what you get when the thing generating the work has been handed examples of good instead of instructions for good.

Both sides of the AI argument are wrong

The AI-will-do-brand crowd. Sure it will, but it won’t do it well without clear guidance and guardrails, and bad branding at the start just means inconsistent, mis-messaged content at scale. There’s a bigger problem coming too. You can ask Claude or GPT to build you a brand, but when 80% of new businesses do exactly that we end up with more same-y brands than we had before AI. Every one built from the same training data and the same handful of frameworks the model knows.

The AI-resistant crowd are wrong in the other direction. It can do brand, and it knows brand theory perfectly well. What it can’t do yet is taste.

Where it works is when you take your strategy and put it into a system the AI can use. The thinking stays yours, and the AI becomes your strategist, assembler and auditor. Give it a system it can work with and your job moves from production and enforcement to creative direction and strategy.

That’s the real prize. Brand stops being the bottleneck and starts being the enabler.

What a rule looks like once a machine has to check it

Our guidelines said the usual things. This is the voice. This is how we want to sound. Active, not passive. All perfectly reasonable, and entirely sufficient for a copywriter who already knows how to write.

Hand the same thing to a model and it does follow the rules. It writes actively. Then it introduces everything the guideline never thought to mention, because it never had to. Every general AI tell. Forbidden words. Non-standard ways of writing our own product name. British spellings where we needed American ones.

The guideline wasn’t wrong. It was written for a reader who turns up with defaults, and a copywriter’s defaults are not a model’s defaults. Nobody writing a brand guideline five years ago thought to specify how many exclamation marks were too many, because no human was ever going to make that mistake.

So the fix isn’t a better sentence. It’s a check.

brand_lint reads the copy and runs it against forbidden phrases, discouraged terms, AI tell words and AI tell patterns, plus British spellings where we need US English. Then the structural rules, sentence construction, exclamation mark counts and a handful of others. Then logic that works out what’s fine and what needs changing, scores it and hands back a grade.

The grade is the part that matters, because it isn’t a gate that only says no. The model takes the score, rewrites what failed and runs again, and if it passes, it ships.

Same on the design side. Colour tokens, spacing scale, typography, checked against what actually got produced. Which catches the more interesting failure. Output that passes because every colour is a brand colour, but leans far too heavily on one of them, or puts the brand colour somewhere it was never meant to go.

Technically that passes. From a brand point of view it doesn’t. That gap is the whole job.

Accessibility and compliance checks sit on top, because those aren’t negotiable and no human reviewer catches them reliably once the volume goes up.

What to systematise, and what to leave alone

This is where I’d push back on my own argument, because it’s the bit people get wrong in the other direction.

Make creative too much of a system and you lose the freedom that produces anything with taste in it. Systematise it completely and it’ll be perfectly consistent and it will not be different, which puts you back at everything looking the same, just by a more expensive route.

There’s a line, and it isn’t subtle once you look for it. Spacing, typography, component structure, the whole design system layer, those should be locked down. Ad creative and campaign work stay freeform, because they need fresh ideas and a rule set that’s too specific leaves no room when creativity is actually required.

Get that boundary wrong and you’ve automated your way to mediocrity. Which is also what happens when you outsource brand entirely, as we found out twice.

Creativity plus AI is scale. AI on its own is volume, and those aren’t the same thing.

If you own a brand, a design system or a product surface, here’s the Monday version

You’ve probably already done most of the hard part and don’t realise it.

You’ve got components in Figma, patterns in Storybook, a tone of voice doc, a positioning deck. It’s all written down somewhere. So how do you connect it to an AI? How do you tell it this is this component, use it for this?

And the harder half, how do you give it the strategy and the thinking behind it? A book of components is fine for a designer, who knows how to use it. So is a tone of voice page, for a writer who already knows how to write. Give either to a model with no context, no strategy and no understanding of why it should reach for that component in that situation, and it won’t use it correctly. It’ll produce something structurally valid and strategically meaningless.

So two questions worth sitting with. How would an AI actually consume this? And what context and strategic thinking does it need that a designer supplies from their own head without noticing?

Answer those, distil it down, hand it over. That’s where content at scale stops being a threat and starts being the point.

---

I build sidenote.ink, which exposes brand systems as an MCP and API so agents and humans compose against the same governed source. It exists because of the argument above.