The short answer
A Taste Profile is the source of truth for your brand: tokens, narrative, rules. A Figma library is the canvas where designers compose with those tokens. You don’t maintain both as parallel systems. You maintain the Taste Profile and export tokens into Figma when designers need them. One source, one update flow, every tool downstream stays in sync.
Capability by capability
| Capability | Figma library | Taste Profile |
|---|---|---|
| Read by Claude Design, Cursor, ChatGPT, Lovable | No | Yes (DESIGN.md, tokens.json) |
| Designers can compose visually on a canvas | Yes | No (export tokens to Figma) |
| Carries brand narrative and rules ("why this blue") | Partially (component descriptions) | Yes (DESIGN.md prose section) |
| Versioned in git with a CHANGELOG | No (Figma version history only) | Yes |
| Lives at a public URL, browsable by anyone on your team | No (Figma file access required) | Yes (yourbrand.tasteprofile.io) |
| Canonical tokens source (W3C DTCG) | Plugin-dependent (Tokens Studio, etc.) | Yes (tokens.json, tokens.css) |
| Engineers consume tokens directly | Limited (Figma dev mode + CSS extraction) | Yes (Tailwind, Bootstrap, plain CSS) |
| Auto-loads into AI tool sessions | No | Yes (SKILL.md for Claude, instructions for others) |
| Cost to build from scratch | Days to weeks of designer time | $999 intro (done-for-you) |
| Cost to maintain after launch | Ongoing designer hours | Edit a markdown file, or ask us |
What designers still do in Figma
Figma is not going anywhere on your team, and a Taste Profile doesn’t pretend to replace it. There’s a clear split of responsibilities once both exist.
Designers still use Figma for the things Figma is built for: composing layouts on a canvas, exploring variants, prototyping interactions, presenting work to stakeholders, and handing off per-component specifics to engineering. Auto-layout, dev mode, and the comment workflow are mature, fast, and visual. None of that is content a Markdown file is competing for.
What the Taste Profile owns instead is the upstream work. The token values, the brand reasoning, the rules about when to use which color, the prose explaining why the headline is light-weight serif instead of bold sans. That layer feeds into Figma (via Tokens Studio or the native W3C variable import), into your codebase, and into Claude Design or Cursor or whatever AI tool a teammate opens this afternoon. One source, many consumers.
Where a Figma library falls short for AI tools
Every AI tool that generates UI today, from Claude Design to Cursor to ChatGPT to v0, reads text. Markdown, JSON, YAML, plain prose. They don’t open Figma files. They can’t walk a tree of frames and components. They can’t parse the description on a Figma variable.
That has four practical consequences:
- The narrative layer is invisible. The reasoning behind your color choices, your “nevers”, your voice rules, none of it is reachable. AI tools can only act on what’s in their context window, and Figma can’t be in their context window.
- Tokens require an export step. Figma variables are great inside Figma. Outside Figma, they only exist if a plugin pushes them somewhere. If that pipeline isn’t set up (or breaks, which it does), AI tools default to shadcn neutrals.
- Updates don’t propagate. Change a color in your Figma library and your AI tools won’t know unless someone rebuilds the export. Most teams forget. Drift sets in within a sprint.
- Every new chat starts blank. No Figma plugin loads brand context into Claude Design or Cursor at session start. There’s no integration that reaches into Figma to fetch your tokens for the next prompt.
A Taste Profile fixes all four because it’s the format AI tools actually consume. Markdown narrative, W3C DTCG tokens, Skills for Claude. The pipeline is the artifact.
How tokens flow once you have both
The Taste Profile is upstream. Everything else consumes from it.
Concretely, the flow looks like this:
- Tokens are defined once in the Taste Profile (W3C DTCG
tokens.json). - Designers pull them into Figma via the Tokens Studio plugin or Figma’s native variable import. Figma is now in sync.
- Engineers consume
tokens.css, the Tailwind config, ortokens.jsondirectly from the repo or the hosted viewer. - AI tools (Claude Design, Cursor, ChatGPT, Lovable, v0) read
SKILL.mdandDESIGN.mdat session start. They generate output that matches what your designers and engineers are already building.
When the brand evolves, you change the Taste Profile. Everything downstream picks it up the next time it’s invoked. No parallel systems to keep in sync. No “wait, which version of the brand is in the Figma file again?”
Useful next read: Why not just paste your brand guide into Claude? covers the longer essay version of this argument, including why the prose layer matters as much as the tokens.
Pricing
$999 intro, done-for-you.
Brand capture from your existing assets, tokens, DESIGN.md, SKILL.md, themed components, hosted viewer, and platform-specific install instructions. Ships in days. Add a Brand Discovery Workshop, Extended Components, or Month of Guidance if you need them.
FAQ
Questions about Figma + Taste Profile
For tokens and brand source-of-truth: yes. For the canvas where designers actually compose mocks, prototype flows, and hand off to engineering: no. Figma is still the right tool for visual design work. The Taste Profile is the upstream spec that feeds tokens and brand context to Figma, to your engineers, and to every AI tool you use.
Two paths today. Export your tokens.json from the Taste Profile and import it into Figma using the Tokens Studio plugin, or use Figma’s native variable import (which supports the W3C DTCG token format we ship). Either way, Figma stays downstream of your Taste Profile.
A one-click export from your hosted Taste Profile is on our short list. Email us if you need it now and we’ll build it into your delivery.
No. We extract from your existing Figma file (or guidelines, or website) as part of the brand capture step. After delivery, your Taste Profile becomes the upstream source of truth. You can re-import the canonical tokens back into Figma when you want them aligned, but you do not need to throw away what you already have.
For tokens and brand rules: yes. The Taste Profile ships tokens.json, tokens.css, and Tailwind and Bootstrap configs that engineers consume directly without ever opening Figma. For component-level CSS handoff on complex layouts, Figma dev mode is still useful. They are complementary: the Taste Profile gives engineers the tokens and rules; Figma gives them per-component layout details when needed.
Build your DESIGN.md
Ready to get AI tools to follow your taste?
Stop getting generic design. Start getting work that actually feels like you without having to tell it each time.