This is the second of two posts on how WRI’s Product Studio uses AI across the design and development process. Part 1 covers how AI can help teams rapidly create and test rough prototypes, using them to sharpen thinking, surface assumptions and gather better feedback before committing to what gets built. In the first part of this series, we argued that when a new piece of software can be built in an afternoon, the discipline of figuring out what to build matters more, not less. As AI-assisted software development makes it possible to generate interfaces on demand, the challenge is no longer simply producing something quickly but ensuring that everything produced is consistent, accessible and trustworthy. WRI software products help turn complex environmental data into information that can guide decisions on policy, resource management and stakeholder engagement. Quicker development can mean more timely action to help people, nature and climate. WRI’s Product Studio makes that delivery phase faster while maintaining rigor. The foundation for this is WRI’s Design System: an open-source component library that is purpose-built for the kinds of applications WRI develops most frequently. To address WRI’s most common user interface challenges — layered maps with complex legends, responsive data panels and large datasets that need to be navigable by non-technical users — we built a custom system. The WRI Design System is documented on Zeroheight and open on GitHub and was co-developed with our UX design partner Usertopia. It represents the accumulation of our answers to thousands of small questions about how information should be presented, how interactions should feel and how trust should be built with people who rely on our tools. The Design System also ensures that we capture learnings across product teams, so the interface patterns that consistently work in one WRI tool don’t have to be rediscovered in the next. AI-Native, by DesignRecently, we invested in making the Design System AI-native. Working with WRI’s Engineering team, Product Studio built context files and Model Context Protocol (MCP) integrations that give AI coding assistants like GitHub Copilot and Cursor accurate knowledge of WRI’s component library. When we ask an AI assistant to build a map with a filterable legend, it will, without this context, invent one from scratch — a generic control that behaves inconsistently, and looks and feels different from every other map in our portfolio. Combining the Design System with AI-assisted development allows the assistant to use WRI’s own legend component instead, with the right design tokens already applied and the interaction patterns we’ve already validated. You can see an example of these building blocks in our live map interface template. The impact on delivery is significant. Even before AI, the Design System let designers produce handoff-ready prototypes roughly three times faster and engineers implement them around 20% faster. These figures are based on team estimates across our recent builds. With AI-assisted development layered on top, teams can now take a geospatial web application from validated concept to a fully functional working interface in hours rather than weeks. Faster Development Changes How Teams Spend Their Time Speed is helpful, but it’s worth being clear about what it does and doesn’t change. You’ll often hear WRI Product teams say ‘the last 10% of any product still takes something like 90% of the effort’ — they’re referring to the quality assurance, accessibility and data integrity checks that separate a nice-looking tool from a trustworthy one. The AI-native Design System doesn’t change that. What changes is how the team spends its time. AI reduces the repetitive work of assembling a consistent interface from known parts so that far more of the team’s attention can be spent on ensuring a tool is trustworthy and delivers value. Team members go from being busy builders repeatedly assembling the same structures to being drivers, conductors and quality assurers directing the work, maintaining standards and judging whether what’s been produced is truly ready for users. Adoption Among Our Partners One sign that WRI’s Design System is becoming a shared standard is that it now guides how partners as well as our own team build WRI applications. Vizonomy, WRI’s technical partner on TerraMatch — WRI’s platform supporting land restoration initiatives — is now applying the WRI Design System in the platform’s redesign. Another example is Blue Coding on Cool Cities Lab. For us, the clearest signal of success is that the same standards for quality and consistency hold as more products are built, whether in-house or by a partner.Standards as a Condition for Responsible AI The nature of WRI’s work requires that every component, interaction and edge case be thoroughly considered before it reaches decision-makers. The Design System makes it possible to move fast in the development phase, without compromising those standards. It speeds implementation and developer handoff while supporting visual consistency. It simultaneously keeps teams anchored to decisions that have already been validated — what the components should do, how they should behave and why. Taken together, the two parts of this series describe the same principle applied at both ends of the process. In discovery, AI raises the stakes by making it easier to build something before you know whether it is the right thing to build. In development, it increased the amount of output that must meet our standards. Discovery helps us determine what to build, while the Design System helps us build it quickly without sacrificing quality or consistency. Together, they show how WRI is adapting its existing practices to take advantage of AI across a product’s design and development process.