Dylan Field on Figma Make + Claude Sonnet 4.5: the era of the 10x designer is just beginning

Summary

Dylan Field, CEO of Figma, spent the weekend stress-testing Claude Sonnet 4.5 inside Figma Make, and came away convinced the model represents a meaningful step-change rather than an incremental update. His focus was the round-trip design workflow: Figma recently shipped a feature allowing users to copy output from Make directly into Figma Design, layering on top of the existing path from Figma Design into Make. Field says Sonnet 4.5 dramatically improved the consistency of that loop, particularly on planning, code-base comprehension, and prompt evaluation across both short and long-running sessions. The timing was coincidental; the feature shipped before Anthropic's release date was known.

Field's model-selection philosophy is blunt: Figma will use whatever model produces the best results, regardless of board relationships. Mike Krieger of Anthropic sits on Figma's board, but Field frames that as irrelevant to procurement decisions. On open-source models, he says Figma watches the space closely but offered no specific commitments.

The 10x Designer Thesis

Field's central argument is that AI is not compressing designer headcount but amplifying designer leverage. He frames this as an emerging "10x designer" era where designers who hold aesthetic craft, UX thinking, engineering awareness, brand strategy, and systems logic simultaneously will become decisive competitive assets.

• Figma's platform is roughly one-third designers, two-thirds non-designers
• Design hiring at enterprise customers is accelerating, not contracting
• Field draws a parallel to the first wave of design acquisitions, citing T. Hannon Lax and John Wax being absorbed into what became Meta's Reality Labs via agency acquisitions as a likely repeating pattern
• Shopify recently acquired a design studio, consistent with this trend

Field invokes Jevons Paradox explicitly: as AI lowers the cost of execution, demand for designers who can generate novel, high-craft output rises rather than falls. Figma itself is hiring aggressively across nearly all functions heading into the next fiscal year.

Figma Make and MCP Momentum

Figma recently upgraded its Model Context Protocol (MCP) offering, moving from a local-only desktop implementation to a cloud service and extending MCP support to Make. The practical effect is that users can connect external coding environments directly to Make's output, eliminating the need to manually export code. Coinbase is cited by name as a customer reporting meaningful workflow improvements from MCP integration.

Field positions MCP as strategic infrastructure for Figma's broader ecosystem play: design context created inside Figma becomes a portable, high-value input for any AI surface consuming inference. The framing is less about locking users into Make and more about making Figma's design data the canonical source of truth across the AI toolchain.

Hardware and the MS-DOS Analogy

On hardware, Field says Figma's primary lens is "what screens do our users need to design for." He is tracking Meta's AR glasses and Neural Band closely, noting he received an early demo roughly 18 months ago and a revised version just before Figma's Config conference. He describes the Neural Band as underappreciated by the market.

His broader take on AI interfaces is that the industry is still in what he calls the "MS-DOS era of AI" — natural language prompts are the current compass for navigating a vast latent space, and the design challenge of building better interfaces for that navigation remains almost entirely unsolved. He sees the proliferation of new form factors, agent protocols, and MCP-style connectors as the beginning of that interface design problem rather than its resolution.

Field's headcount and product investment signals suggest Figma is positioning for a sustained expansion cycle, betting that design as a discipline moves decisively up the value stack as AI commoditizes software execution.