Figma's AI chief: 'Humans bring judgment, empathy, and taste'

David Kossnick, Figma’s Head of AI Products, doesn’t mince words about who’s really in charge of the creative process. ‘Humans bring judgment, empathy, and taste—qualities that make them the true pilot, not just the co-pilot,’ he said in a recent interview. That perspective is shaping how the design giant embeds AI across its entire platform, from auto-renaming layers to generating production-ready code.

The company’s approach treats AI as both a platform shift and a core capability, not a flashy add-on. Kossnick points to Figma Make, a prompt-to-app tool that spits out production-grade code from natural language, images, or structured frames. It’s built for coders and non-coders alike—PMs, marketers, and designers who typically hit technical walls during prototyping. What sets it apart, he argues, is what happens after the AI does its work. Unlike tools that leave you stuck with whatever the model produces, Figma lets users tear into every layer—visual, language, and code—to refine things with full control.

That philosophy extends to how teams actually work together. Figma’s multiplayer DNA means two people can co-create with an AI assistant in the same file, watching each other’s avatars move in real time. Even image generation in FigJam and Slides has become a collaborative ritual, with teams remixing brand-aligned visuals side-by-side. Kossnick describes a company tradition where teammates use OpenAI’s image editing to create playful anniversary cards from each other’s avatars—a small but telling example of how AI tools can foster connection rather than isolation.

The developer handoff pipeline is getting a similar rethink. Dev Mode already streamlined things with structured CSS and tokens. Now Figma’s MCP server lets developers invoke a coding agent that translates mockups into production-ready code with full context—no copying and pasting required. Kossnick notes that designers are getting so precise with their prompts in Figma Make that engineers are starting to treat the output as a legitimate handoff artifact. It’s a signal that the line between prototype and production is blurring faster than most design teams are prepared for.