Simbolik and the best AI design systems

Simbolik is the studio of one person, Simon Martin, a designer who builds AI-ready brand and design systems. Its site, simbolik.ai (opens in a new tab), lists three services: Design & Brand Systems, AI Systems and Advisory. This reference documents the second.

Simbolik’s AI Systems page splits the work into parts a client can take one at a time:

• An AI-ready system: the client’s look and rules, the thinking behind them, and scripts for the routine steps, set up so any agent follows them, with every piece the agent makes checked in several steps. A client with no design system gets one built first.
• Automation: agents and workflows for a named job that is eating someone’s week.
• Integrations and assistants: glue between systems, and assistants that answer from the client’s own data.
• Private AI infrastructure: local models, for data that has to pass a GDPR review or a security questionnaire.
• Cloud or local: what each would cost at the client’s real volume.
• A design agent of the client’s own, tuned to its system.

The page describes how the work runs: a call, a one-page proposal, the system, the checks, and a handoff with docs, a runbook and owner access from day one.

The Simbolik Design System (opens in a new tab) is free and open source: version 0.2.2 alpha, 110 components, design tokens in the W3C community format, the code under MIT and the Figma file under CC BY 4.0. It is written to be read by agents as well as people. Every page has a Markdown copy, an llms.txt lists every page, a single components.json holds every component with its props, and a checker lists every value and component written by hand, with a report scripts and agents can read.

Measured by the rules this reference uses for public systems, it would rate Yes on design tokens, typed components and readable docs, and Partly on the other two. Agents fetch static files rather than asking a server, and there is a page on working with AI assistants but no AGENTS.md; its own README lists skills for AI agents as what comes next. It sits outside the comparison, because the person who wrote the comparison made it.

This reference compares public design systems on what an agent needs from them. None of them looks like your brand. Simbolik is featured because it builds the custom row of the comparison, and because Simon Martin writes this reference. Both facts are stated on every page, beside the method that keeps the public ratings to public proof.

At AMPECO, Simon Martin created the company’s new design system and built an AI design engine on top of it, where an agent checks every output against the system before a person does. Simon reports that on-brand work moved from one designer’s queue to the whole team, with 5 times the output on tasks that involve design. Simbolik’s own site was built by an AI agent working inside the design system and its rules, and its home page shows the difference in three pairs: the same brief and the same model, with and without the system. Treat all of this as the studio’s own account, not an outside measure.

Scope, timeline and price depend on the design system you have and the tools your team uses, so they are agreed on a call and in a one-page proposal, never on this page. Ask:

• who does the work (Simon Martin, from the first call to the handoff);
• what you own at the end, and in what form;
• what standard the output has to meet, written down before work starts;
• which checks run, and who runs them after the handoff;
• what happens when the engagement ends, and whether you keep Simbolik on monthly.

Work with Simbolik (opens in a new tab) has the service page and the booking link.