Figma co-founder Dylan Field thinks MCP is not a perfect protocol

Figma announced 20 new partners for the open-source Model Context Protocol (MCP), expanding integrations while co-founder Dylan Field said the protocol is imperfect and leaves room for improvement. Figma’s CEO cited abuse risks from exposing more company data through MCP; Field also said Figma plans to support Pi.

Analysis

The strategic tension is value capture: broader agent connectivity can make Figma more useful as a system of record, but may shift the user interface—and bargaining power—to model vendors. If agents let users create or modify design work without spending time inside Figma, integrations could support retention while weakening seat growth or pricing power. These effects are not established by partner count; usage, paid conversion, and workflow completion matter more.

Near term, the announcement is not enough to justify a directional FIG trade. Over the next 1–3 months, watch for evidence of active usage, enterprise controls, and whether integrations drive incremental paid adoption rather than merely exposing data. A breach or misuse involving connected data could turn openness into a trust and compliance liability, with costs potentially extending beyond the affected workflow. Over 6–18 months, the key question is whether Figma remains the authoritative design workspace or becomes a replaceable data source behind frontier-model interfaces.

The contrarian angle: protocol openness is not automatically a win for the platform adopting it. Frontier labs may capture user interaction and learn from context while Figma bears security, support, and governance burdens. Conversely, restricting access too aggressively risks ceding developer mindshare and workflows to competing design tools. Treat partner breadth as optionality, not proof of monetization.

Key Decisions for Investors

• No event-driven FIG position on this announcement alone; partner count is a weak proxy for revenue or retention impact.
• Set a 1–3 month diligence trigger: seek evidence on MCP-driven active usage, paid-seat conversion or retention, enterprise adoption, and controls over what data agents can access. Reassess only if usage translates into measurable economics.
• Monitor for security or privacy incidents involving connected design data, and for changes in enterprise guidance or customer churn. Such evidence would strengthen the view that integration costs and trust risk outweigh ecosystem benefits.
• For a conditional longer-horizon thesis, favor FIG only if agent integrations increase workflow depth and paid retention without weakening seat economics; falsify it if management indicates agent usage is substituting for paid seats or if design work migrates to model-led interfaces.