Two ways in. Each step de-risks the next.

A short, bounded engagement to find where your substrate (the design system, tokens and pipeline underneath your product) is broken and what it's costing the business. I get into the product, the codebase, the Figma files, and whatever brand or digital guidelines exist, and I talk to the developers carrying the design load. I find the pressure points: duplicated components, broken handoff, token chaos, design calls made in silos. Then I trace what they're costing you in velocity and defects.

It ends with a report installed in your repository. The checks read your code and your token source: values hardcoded past the system, semantic tokens that can't flip between light and dark or between brands, theme keys your build never fills. It runs on your machine and needs nothing from me, so you can run it again in three months and see whether anything improved.

You also get the diagnosis: where the substrate is broken, what it's costing, and what fixing it would involve, scoped and priced. Both are yours whether or not you go further. And the Build, if it happens, is scoped from evidence rather than a sales call.

The other way in, for a product that doesn't exist yet. There's nothing to audit, so this one earns its scope by building. I take the brand foundations, or whatever exists of them, and author the token layer properly: DTCG source, primitives and semantics, light and dark resolved and checked. It's wired into whatever your team is shipping on, which for most new products means shadcn and Tailwind. Then the sync, the build step, and the guardrails that stop it drifting. It ends with one real screen shipped from the system, not a page of swatches.

If you already have shadcn, good. I build on it rather than replacing it. It gives you semantic variables with a dark block over the top; what it hasn't got is the layer above. No authored source, no build, no sync, no guardrail. So you have the names without the ownership, which holds until the first rebrand or the second developer, at which point there's nothing upstream to change and forty files to edit by hand. You get a token layer you own, connected to the code you ship: one place to change a colour, a build that fails loudly when someone bypasses the system, and a scoped, priced Build if you want the rest.

The design system, built and connected. The working thing. Tokens, a component library, page patterns, and the pipeline connecting Figma to the codebase: Style Dictionary, GitHub, whatever the working thing requires. Built iteratively, in your workflow, with your developers, not delivered over a wall at the end.

You get one source of truth your developers ship from: fewer duplicated components, fewer design calls made in silos, less defect debt, and a system that lives in your codebase rather than a Figma file someone has to translate. The exact figure is set by the Diagnostic, usually £12,000–£18,000 depending on what it finds. I don't price unscoped work, so there's no Build without a Diagnostic first.

Orin stays close and the system keeps living: roughly three to four days of attention a month. Iterating the system as the product grows, reviewing new components before they fragment, keeping the pipeline healthy, and being the design brain your developers can call on when they're making a decision they're not sure about.

The system doesn't decay back into an artefact. New features ship consistent with the system, and the design load your developers carry stays light. It's a fraction of what one junior designer costs, and it's cancellable with 30 days' notice. No lock-in.

Some work isn't mine to do, and saying so early saves everyone time. I don't price a Build without a Diagnostic first, and I don't scope a system to fit a "can you do it in forty hours" budget. That treats the system as a commodity, and it isn't one. Brand guidelines, marketing surfaces, and social media assets are legitimate needs, but they aren't what I do.

I build; advice without building isn't the offer, so strategy-deck-only engagements aren't a fit. And Orin is a partner, not a vendor. If that's the shape you're after, I'll happily point you elsewhere.