[
  {
    "id": "agent-ready-design-systems-index",
    "title": "Agent-Ready Design Systems Index",
    "author": "Kiryl Zhukouski, DesignSystems.one",
    "published": "2026-06-10",
    "kind": "study",
    "url": "https://www.designsystems.one/ai-ready/systems",
    "description": "Scores a field of 37 against five first-party signals: an MCP server, llms.txt, DTCG tokens, an installable component registry, and published Code Connect mappings. Wrappers and third-party shims do not count. Re-audited quarterly, and downloadable as JSON and CSV under CC BY 4.0.",
    "added_on": "2026-07-28"
  },
  {
    "id": "building-design-system-components-with-agent-teams",
    "title": "Building design system components with agent teams",
    "author": "Kaelig Deloumeau-Prigent",
    "published": "2026-04-22",
    "kind": "study",
    "url": "https://www.kaelig.fr/design-system-components-with-ai-agent-teams/",
    "description": "A build report: eight agents across an Understand, Build and Verify pipeline turning a Figma file into a production Menu component, each with a named artifact, exit criteria and a retry budget. Full autonomy hit a ceiling, and the fix was a human gate rather than a better prompt.",
    "quote": "The pipeline could follow rules but it couldn't question.",
    "added_on": "2026-07-28"
  },
  {
    "id": "zeroheight-design-systems-report-2026",
    "title": "Design Systems Report 2026",
    "author": "zeroheight",
    "kind": "study",
    "url": "https://report.zeroheight.com/",
    "description": "A practitioner survey, n=147. One in ten teams has AI in the process, just under half are experimenting, and the rest are not. Code generation leads the use cases, then documentation generation, then prototyping. Where this list mostly reads artifacts, this one asks people.",
    "added_on": "2026-07-28"
  },
  {
    "id": "ai-ready-design-system-roadmap",
    "title": "AI-ready Design System Roadmap",
    "author": "Ilya Greben, Design Systems Surf",
    "kind": "tool",
    "url": "https://designsystems.surf/products/ai-ready-roadmap",
    "description": "An 80-page PDF that walks a system through 20 areas across four stages — Define, Create, Adopt, Evolve. Each area carries checkpoints, the context an AI tool would need to act on it, and a readiness test for whether an agent can apply the decision without filling in the missing logic itself. Drawn from the systems Design Systems Surf has cataloged since 2023. Free, behind a form that asks for an email.",
    "quote": "An AI-ready design system makes its decisions, semantics, rules, relationships, and constraints explicit enough for AI tools to apply them without inventing missing logic.",
    "added_on": "2026-08-06"
  },
  {
    "id": "zeroheight-design-system-maturity-model",
    "title": "Design System Maturity Model",
    "author": "zeroheight",
    "kind": "tool",
    "url": "https://zeroheight.com/maturity/",
    "description": "A free assessment that places a system on six independent axes rather than one score, each running through four stages. One of the axes is AI readiness, which makes this the closest thing to a self-serve version of the question this report asks. Built on the four-stage model from Sparkbox.",
    "quote": "Whether your design system is the default context that AI tools build from, or whether AI is silently working around it.",
    "added_on": "2026-07-28"
  },
  {
    "id": "vercel-design-md",
    "title": "How our agents build on-brand pages with design.md",
    "author": "John Phamous, Vercel",
    "published": "2026-08-31",
    "kind": "essay",
    "url": "https://vercel.com/blog/how-our-agents-build-on-brand-pages-with-design-md",
    "description": "What Vercel gives agents working outside its repositories, where the `product-design` skill and the components it points at are not there to read. Porting that skill into a public prompt failed, because every model interpreted the same description of the visual language differently, so `design.md` was rewritten from scratch against seven frozen scenarios and rerun on Claude Opus 4.8 and Codex with GPT-5.5 after each change. Three layers came out of it: prose judgment in the file, a public stylesheet the agent never reads because it loads at render time, and deterministic checks for the mechanical failures. Over 200 runs went in. A six-page comparison found 39 known failures with the file against 91 without, 57% fewer, and states its own caveats: the checks catch only failures already named, six pages are too few to claim quality, and every page still had at least one failure serious enough to block shipping.",
    "quote": "Every line of guidance in design.md earned its place through the eval loop.",
    "added_on": "2026-09-01"
  },
  {
    "id": "bill-design-system-bilingual",
    "title": "How We Made Our Design System Bilingual: One Source of Truth for Humans and AI Agents",
    "author": "David Cai, BILL",
    "published": "2026-08-17",
    "kind": "essay",
    "url": "https://www.bill.com/engineering/how-we-made-our-design-system-bilingual",
    "description": "A case study on making Trinity, BILL’s design system, usable by AI agents without writing a second set of docs. Storybook stays the source of truth, and three things sit on top of it: a CLI that exposes component APIs and design tokens as structured JSON, a hosted MCP server for agent access, and agent skills that enforce compliance. The point is to ground an agent in verified component knowledge so it stops inventing hardcoded hex values or the wrong component. A/B tests are reported to cut token usage by 31–63% per task.",
    "quote": "Our bet: instead of writing new docs for AI, turn the existing source of truth into a shared interface for humans and agents.",
    "added_on": "2026-08-24"
  },
  {
    "id": "uber-base-mcp",
    "title": "Base MCP: Teaching AI agents to speak the Base Design System",
    "author": "Ali Asghar Aamir, Uber Design",
    "published": "2026-08-12",
    "kind": "essay",
    "url": "https://uber.design/base-mcp",
    "description": "A write-up of Base MCP, Uber’s internal remote server that connects AI agents to the Base design system so an agent queries current documentation instead of guessing. It exposes 280 documentation pages, 700+ design tokens with light and dark variants, and 3,200+ keyword-searchable icons in filled and outlined variants, alongside pointers to each platform’s component implementations, with nothing to install locally and no API keys to manage. Reports 500+ users and 50,000+ tool calls since launch.",
    "quote": "Base MCP is an internal tool built around one idea: design systems should be machine-readable.",
    "added_on": "2026-08-24"
  },
  {
    "id": "use-ai-to-need-less-ai",
    "title": "Use AI to Need Less AI",
    "author": "TJ Pitre, Southleft",
    "published": "2026-07-26",
    "kind": "essay",
    "url": "https://southleft.substack.com/p/use-ai-to-need-less-ai",
    "description": "Splits the work into authoring, which is judgment-heavy and happens once, and enforcement, which recurs forever and has to be boring, then surveys what is being built on that split: the DTCG token format, deterministic spec extraction from Figma, a documentation schema, and a set of contract layers that gate a generated component against a machine-readable definition. Discloses that one of them is the author’s own.",
    "quote": "A model can be talked around. A schema can't.",
    "added_on": "2026-07-28"
  },
  {
    "id": "design-systems-need-evals",
    "title": "Design systems need evals",
    "author": "Murphy Trueman",
    "published": "2026-07-17",
    "kind": "essay",
    "url": "https://blog.murphytrueman.com/design-systems-need-evals/",
    "description": "Argues that a design system with agents pointed at it should ship a regression suite the way it ships a changelog. Your instructions tell an agent what the rules are; only running it under real conditions tells you whether it followed them.",
    "quote": "You've taught agents your design system. You still don't know whether they're following it.",
    "added_on": "2026-07-28"
  },
  {
    "id": "life-after-the-prototype",
    "title": "Life After the Prototype",
    "author": "TJ Pitre, Southleft",
    "published": "2026-06-22",
    "kind": "essay",
    "url": "https://southleft.substack.com/p/life-after-the-prototype",
    "description": "A sequence for turning a generated prototype into code an engineer will take: a design branch with no path to production, the model pointed at real tokens and components over MCP, the rules written in a file the agent reads every time, generation into Storybook so there is nothing left to invent, then asking the model where it deviated and what it is least confident about. What it reports doubles as an audit of where the system is thin.",
    "quote": "The prototype was never the hard part. The handshake is.",
    "added_on": "2026-07-28"
  },
  {
    "id": "my-beef-with-agentic-design-systems",
    "title": "My beef with agentic design systems",
    "author": "TJ Pitre, Southleft",
    "published": "2026-06-19",
    "kind": "essay",
    "url": "https://southleft.substack.com/p/my-beef-with-agentic-design-systems",
    "description": "Separates the architecture question from the accountability one. Agents working in a loop is a claim about capability; a design system is a way of encoding collective judgment and holding people to it. Offers one test for anyone selling the former as the latter.",
    "quote": "what rejects the agent's output, and who decided the rule it's being rejected against?",
    "added_on": "2026-07-28"
  },
  {
    "id": "context-based-design-systems-revisited",
    "title": "Context-Based Design Systems, Revisited",
    "author": "Southleft",
    "published": "2026-05-12",
    "kind": "essay",
    "url": "https://southleft.com/insights/design-systems/context-based-design-systems-revisited/",
    "description": "Returns to that framework a year on and sharpens it against the agentic writing that appeared in between. The disagreement is about the operating model: propose-and-approve puts the agent upstream and the human at the gate, where this keeps the person upstream setting the constraints the agent executes inside. The change it argues actually mattered is designers opening the pull request.",
    "quote": "The orchestrator is not the reviewer at the gate. The orchestrator is the one driving.",
    "added_on": "2026-07-28"
  },
  {
    "id": "agentic-ai-design-systems-and-figma",
    "title": "Agentic AI, design systems and Figma",
    "author": "Christine Vallaure",
    "published": "2026-03-31",
    "kind": "essay",
    "url": "https://christinevallaure.substack.com/p/agentic-ai-design-systems-and-figma",
    "description": "The design side of the same shift. Naming conventions, semantic tokens and complete component states stop being documentation hygiene and become the instructions a machine follows, which puts more of the outcome in the designer’s hands and less of it up for interpretation.",
    "added_on": "2026-07-28"
  },
  {
    "id": "ai-in-design-systems-whats-changing-in-2026",
    "title": "AI in design systems: What’s changing in 2026",
    "author": "Elyse Holladay, zeroheight",
    "published": "2026-03-25",
    "kind": "essay",
    "url": "https://zeroheight.com/blog/ai-in-design-systems-whats-changing-in-2026/",
    "description": "Three moves for the year: MCP servers, skills and design guidelines; designing in code rather than on a canvas; and shipping patterns as structured relationships instead of components. Argues a system has to document when and why, not only how, because the why is what a model otherwise guesses at.",
    "quote": "AI is finally starting to change what a design system ships.",
    "added_on": "2026-07-28"
  },
  {
    "id": "in-the-open-what-the-modal-reveals",
    "title": "In the open: What the modal reveals",
    "author": "Murphy Trueman",
    "published": "2026-03-13",
    "kind": "essay",
    "url": "https://blog.murphytrueman.com/in-the-open-what-the-modal-reveals/",
    "description": "Reads the modal side by side in Carbon, Material UI, Polaris and Radix, then shows how each one’s naming and composition choices commit it to a different contract with whoever consumes it, human or agent.",
    "quote": "Where a component expresses its structure is where it communicates its contract.",
    "added_on": "2026-07-28"
  },
  {
    "id": "context-based-design-systems",
    "title": "Context-Based Design Systems: A New Model for the AI-Driven Product Lifecycle",
    "author": "Southleft",
    "published": "2025-07-03",
    "kind": "essay",
    "url": "https://southleft.com/insights/design-systems/context-based-design-systems-a-new-model-for-the-ai-driven-product-lifecycle/",
    "description": "Names a lifecycle where every phase inherits structured meaning from the one before it, so a component travels with what it means, how it behaves and when it applies rather than only what it looks like. Introduces the context engineer, whose job is the fidelity of that chain rather than the prompts.",
    "quote": "What's missing isn't data, it's context.",
    "added_on": "2026-07-28"
  },
  {
    "id": "your-next-design-system-user",
    "title": "Your next design system user",
    "author": "Murphy Trueman",
    "published": "2025-06-04",
    "kind": "essay",
    "url": "https://blog.murphytrueman.com/your-next-design-system-user/",
    "description": "Makes the case for treating a component library as an interface other software consumes: explicit semantic contracts, purpose-driven naming, machine-readable documentation. The earliest work on this list, and the one the others keep arguing with.",
    "quote": "Your design system is already an API; the question is whether it's a good one.",
    "added_on": "2026-07-28"
  },
  {
    "id": "design-systems-for-the-machines",
    "title": "WWC26 - Design Systems for the Machines - How to Make AI Understand Your UI",
    "author": "Jennifer Wjertzoch, on WeAreDevelopers' channel",
    "published": "2026-07-17",
    "kind": "talk",
    "url": "https://www.youtube.com/watch?v=S2Z6jwexHgc",
    "description": "A WWC26 talk building the case that a design system’s real job was always reducing ambiguity, for humans and now for machines too. Walks a demo button through three layers — semantic HTML, ARIA, then a data-* intent attribute and a component manifest — and reruns an accessibility scan and a scripted browser agent against both versions to show what each layer changes. Closes with a four-step plan for a first quarter: audit with an agent, write the conventions into an ADR and ESLint, gate CI on the scan, and test with an agent the way you’d test with a screen reader.",
    "quote": "The future of design systems is not visual consistency. It's reducing ambiguity for humans and machines alike.",
    "added_on": "2026-07-31"
  },
  {
    "id": "agentic-design-systems-in-2026-with-brad-frost",
    "title": "Agentic Design Systems in 2026 with Brad Frost",
    "author": "Brad Frost, on Chromatic's channel",
    "published": "2025-12-11",
    "kind": "talk",
    "url": "https://www.youtube.com/watch?v=Vg78K3t9KYc",
    "description": "An eighty-minute conversation rather than a conference talk, recorded before most of the writing on this list. The oldest work here, and the one most likely to have been overtaken.",
    "quote": "AI is rapidly reshaping who (or what) uses your design system.",
    "added_on": "2026-07-28"
  },
  {
    "id": "ai-and-design-systems-course",
    "title": "AI & Design Systems",
    "author": "Brad Frost, Ian Frost and TJ Pitre",
    "kind": "course",
    "url": "https://aianddesign.systems/",
    "description": "Six chapters covering AI fundamentals, tooling setup, improving an existing system, and rolling the result out across an organization. Released iteratively, with live jam sessions and a Slack community alongside.",
    "price": {
      "amount": "$995",
      "buys": "one license, with 20+ hours of demos and recorded sessions available so far"
    },
    "added_on": "2026-07-28"
  },
  {
    "id": "build-scalable-ui-in-figma-and-ai",
    "title": "Build Scalable UI in Figma & AI — Designers Ready for Agentic Workflows",
    "author": "Christine Vallaure",
    "kind": "course",
    "url": "https://maven.com/moonlearning/figma",
    "description": "A four-week cohort for designers, with no engineering background assumed. The first three weeks are the file structure an agent reads, components, variants, variables, styles, auto layout and naming, and where each one does and does not map to front-end code. The last week is MCP, Code Connect, design markdown files and skills, ending with building a lightweight agent against the project made during the course.",
    "quote": "Skip this, and all AI you add is built on sand and useless.",
    "price": {
      "amount": "$799",
      "buys": "a seat in the cohort, with live sessions, lifetime access to the recordings and the Figma practice files"
    },
    "added_on": "2026-07-28"
  },
  {
    "id": "into-design-systems-course",
    "title": "Design Systems Course",
    "author": "Sil Bormüller, Into Design Systems",
    "kind": "course",
    "url": "https://www.intodesignsystems.com/design-systems-course",
    "description": "Six modules taught by practitioners from WhatsApp, GitHub, Figma, Adobe, Miro, Atlassian and Indeed: agentic design systems, making a system machine-readable, AI infrastructure, and shipping production code as a designer.",
    "price": {
      "amount": "$599",
      "buys": "lifetime access to 15+ hours of video lessons"
    },
    "added_on": "2026-07-28"
  },
  {
    "id": "figma-handoff-to-ai-agents",
    "title": "Figma Handoff to AI Agents",
    "author": "Christine Vallaure, moonlearning",
    "kind": "course",
    "url": "https://www.moonlearning.io/agenticdesign",
    "description": "A self-paced course on getting a Figma file ready for an agent to build from. It opens on what an agent is and what MCP does, then walks variables, styles, components and auto layout through to tokens and CSS, saying what to change in each so a machine reads it the way it was meant. It closes on the context added on top: design markdown, skills and layer mapping.",
    "quote": "Something that started in enterprise design systems working with AI is now trickling down fast, and it changes what your Figma file is actually for.",
    "price": {
      "amount": "29€ a month, or 229€ a year",
      "buys": "membership access to this course, every other released one, and the Figma source files"
    },
    "added_on": "2026-07-28"
  }
]