Design systems: From fundamentals to AI-era foundations

Design systems remain the connective tissue between design and production, delivering consistency as organizations scale. With AI accelerating workflows, the need for a codified foundation is becoming more pronounced—AI tools require structured context to produce output that aligns with design intent rather than introducing inconsistencies.

As Schema by Figma returns for its 2025 edition, the resources below offer a path forward, whether you are building a system from scratch or trying to measure the value of an existing one.

Building a foundation: History and documentation

The concept of systematic design is not new. Its roots lie in graphic design and print media, where style guides and typographic standards first codified visual consistency. As mid-20th-century businesses recognized the value of brand identity, comprehensive guidelines became standard. The digital era pushed those guidelines from paper to pixels, with companies like IBM, Microsoft, Apple, and Google translating them into operating-system conventions that internal teams and third-party developers were expected to follow.

Documentation is what turns abstract principles into a usable shared reference. It gives designers and developers the confidence to adopt the system. At Alaska Airlines, the team building the Auro design system made documentation a top priority, producing guidance on Figma features such as auto layout and merging branches so designers knew exactly where to begin. But effective documentation also has to serve different audiences—some practitioners need visual context, others benefit from high-level guidance, and still others require detailed specifications.

Measuring the value of a design system

Business impact can be tracked through usage metrics, adoption rates, and consistency scores. At athenahealth, Senior UX Designer Veronica Agne used consistency scores to spot an uptick in detachments for a container component. That signal prompted a question worth investigating: whether the cause was a bug, a desired feature enhancement, or unexpected combinations of existing elements—all valuable information for the system's evolution.

Why AI needs design context

AI tools are increasingly able to translate code into designs and convert visual designs into working components. But without context, that output can miss the mark, generating inconsistent artifacts that contradict design guidance. Design systems provide the structure AI agents need, enabling them to produce on-brand results that reflect established practices. The payoff is smoother handoffs, less rework, and consistently higher-quality products—a benefit that grows as teams work to justify the investment in maintaining a system.

In practice: Systems at scale

Spotify's cross-platform sync

When Spotify's product surface expanded to over 2,000 devices, the need was not just coverage but coherence. Design systems manager Juli Sombat's team audited each surface to understand its constraints—desktop had ample real estate, TV demanded consideration of viewing distance, Watch required condensed information. The team aimed for a middle path between perfect consistency and full platform independence, producing a system that met each platform's needs while preserving the Spotify brand.

eBay's centralized playbook

eBay's new design system, Evo, announced in November 2024, modernized the company's 30-year-old guidelines. The resulting "eBay Playbook" spans over 300 pages of documentation designed for internal teams and agency partners alike. Rather than splitting brand and product guidance, the team created a centralized source of truth that ties brand, design system, and code together.

Pinterest's open-sourced adoption metric

Design Technologist Ravi Lingineni wanted to quantify the impact of Pinterest's Gestalt design system. He built a tool called FigStats using Figma's REST API. Running nightly across all of Pinterest's Figma files, it tallies interactions with Gestalt and produces an organization-wide adoption percentage. The Gestalt team open sourced the code so other design systems teams could track adoption similarly.

Further reading

Conversations ahead

These themes—adoption, documentation, AI readiness, and ROI—are among the topics slated for discussion at the Schema 2025 virtual event. Speakers include Figma's Paige Costello and Loredana Crisan, design system expert Nathan Curtis, and Rachel Been of Expedia Group.