cd /news/ai-agents/why-buildkite-hired-evil-martians-fo… · home › topics › ai-agents › article
[ARTICLE · art-164258] src=evilmartians.com ↗ pub= topic=ai-agents verified=true sentiment=↑ positive

Why Buildkite hired Evil Martians for an agentic-first design system

Buildkite hired consultancy Evil Martians to build an agentic-first, framework-agnostic design system and component library, including a system of AI skills that gives coding agents access to component APIs, patterns, accessibility requirements and design intent. Evil Martians delivered a platform UI and frontend audit, a shared token and styling foundation with light and dark theme support, initial reusable components documented in Lookbook, Figma explorations, accessibility and interaction guidance, machine-readable component documentation, and multiple production pull requests supporting both Rails and React. The engagement targets a specific failure mode: an agent that cannot see the component API and design intent guesses, and each guess becomes a new one-off component the next session builds on.

by read3 min views1 publishedNov 4, 2026
Why Buildkite hired Evil Martians for an agentic-first design system
Image: Evilmartians (auto-discovered)

Buildkite is a high-scale, hybrid CI/CD platform that lets engineering teams run fast automated build, test, and deployment pipelines.

Buildkite brought in Evil Martians to create an agentic-first, framework-agnostic design system and component library: a foundation for a more consistent, accessible, and faster interface that lets product engineers ship new features faster.

This is the story of our engagement, and a question: why would a team of brilliant developers hire consultants at all?

#

Buildkite is a team of brilliant, opinionated developers who love the product they’re building. Their client list makes that clear. Uber, Slack, OpenAI, Anthropic, SpaceX, and many more.

#

Buildkite didn’t need us because they were short on engineers. They wanted external validation of their technical approach to a dual design system implementation, and someone working full time on gathering opinions and running the experiments.

So, while the Buildkite team was busy shipping new features, Martians did thorough, uninterrupted research, presented the options, collected feedback, and built out both the design system and the AI harness on top of it.

Seeing many companies struggle with the same challenges today gives us an advantage. We were able to come up with a solution for Buildkite’s case in a short span of time.

#

As a product grows, so does the need to keep its design consistent across every surface. New screens should use the existing components instead of reinventing unnecessary ones.

That gets harder once coding agents write a share of the frontend. An agent that can’t see the component API, the patterns, and the design intent behind them will guess, and every guess becomes a new one-off component the next session builds on.

#

The work started with an audit of the existing UI and frontend architecture. From there, we developed a shared token and styling foundation, explored a component model that could work across Rails and React, and documented components through Lookbook and machine-readable guidance.

Our engagement produced several deliverables:

  • A complementary system of AI skills that gave coding agents access to component APIs, patterns, accessibility requirements, and design intent, used both to create components and to consume them
  • Design system foundations
  • An updated token architecture with light and dark theme support
  • Initial reusable components and component patterns
  • Figma explorations
  • Accessibility and interaction guidance

The team also advised on modernizing the frontend stack: using modern CSS, native browser APIs, and web standards wherever possible to reduce custom JavaScript and long-term maintenance.

#

Progress was demonstrated through concrete, production-ready outputs rather than traditional business metrics. Evil Martians completed a platform UI and frontend audit, established a shared token and styling foundation with light and dark theme support, introduced the first reusable components into Lookbook, and validated an approach that could support both Rails and React.

We also delivered Figma explorations, accessibility and interaction guidance, machine-readable component documentation, a system of AI skills for coding agents, and multiple production pull requests implementing the proposed architecture.

Together, these provided a working foundation and clear technical direction for a framework-agnostic, agentic-first design system.

And based on this work, we built a framework and explained the technical side of the implementation in AI makes design system guardrails mandatory; this framework delivers them.

#

Corey Ginnivan is Head of Design Engineering at Buildkite: Shipping features faster than your UI can stay consistent, or watching coding agents reinvent components you already have? Evil Martians build design systems for people and AI agents alike, from the audit to the AI skills that keep agents on your components. Talk to us.

── more in #ai-agents 4 stories · sorted by recency
── more on @buildkite 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
→ Live at https://your-agent.zahid.host ✓
Get free account → Pricing
from €0/mo · no card required
LIVE [news/why-buildkite-hired-…] indexed:0 read:3min 2026-11-04 · —