Figma acquires Diagram, betting AI shifts design from pixels to patterns

For years, Figma has watched the community experiment with artificial intelligence through its open API — nearly 100 AI-powered plugins have emerged from users so far. One of the earliest experiments came from Jordan Singer, who in 2020 built a plugin called “Designer” that used GPT-3 to generate design ideas from a simple prompt. Singer went on to found Diagram, a startup building at the intersection of design and AI. Figma has now acquired the company, bringing Singer and the Diagram team (Siddarth, Andrew, Marco, and Vincent) into the fold.

The acquisition fits a strategy Figma has been building toward internally. Over the past year, the company has grown a dedicated machine learning team and accelerated early development of its own AI platform. The core thesis: AI is not a feature but a platform capability that touches every stage of product development, from discovery to production code.

AI’s role across the product lifecycle

Figma sees AI playing a role at each phase of the design and development process:

  • Discovery: AI might generate and synthesize early ideas from a simple prompt.
  • Design: AI could tap into existing designs and design systems, surfacing component recommendations that speed up the move to a first draft.
  • Development: AI could help developers infer context faster and generate better production code.

The goal is not to automate the job but to lift it. By handling the mechanics of execution, AI frees designers to spend more time on problem solving — the reason many entered the field in the first place. As with past shifts in technology, from the printing press to the smartphone, design work evolves rather than disappears. The question is what changes first: how design happens, what gets designed, and who does the designing.

From components to patterns

Design systems already pushed the discipline away from fine-grained pixel work toward reusable components. The next step, in Figma’s view, is moving from those components to patterns. If a login screen today means assembling email fields, password inputs, and buttons, AI might let designers imagine entirely new authentication flows that replace email or Touch ID. It might also suggest a color palette from a color wheel based on the emotional tone of a project. The trajectory mirrors the atomic design methodology popularized by Brad Frost, who has noted that “increasingly, design is curatorial.” That curatorial role — applying taste and directing concepts — becomes more central when AI handles the repetitive assembly.

Rethinking the experiences themselves

AI may also reshape what gets built. Chat-based interfaces like ChatGPT are pushing digital experiences away from the app-based model back toward conversational interaction. The user who wants an Uber to the airport doesn’t want to open an app, set a location, evaluate options, and confirm — they want to say “Get me to JFK.” Product builders will need to ask whether AI can deliver the same or better user experience with fewer clicks and decisions, bridging the gap between intention and action more directly.

Widening the design floor, raising the ceiling

Figma anticipates that product roles will blur further. More people will become visual creators, and existing designers will be able to attempt more ambitious work. A useful mental model: the craft has a ceiling — constrained by tooling — and a floor, the minimum skill needed to participate. AI raises the ceiling, enabling more creative output with more powerful tools, while lowering the floor so anyone can join the visual collaboration. The result is a shared design space, a natural extension of Figma’s real-time collaboration features that have already blurred the boundaries between roles.

There are still many unknowns. But the throughline, Figma argues, is that the core of the work — solving problems — stays constant. The company wants feedback from the community on what to build next, and plans to develop the platform in the open.