Field said the open-source protocol, which enables AI agents to connect with external applications and information, requires substantial improvement. His comments followed Figma’s announcement of 20 additional MCP partners at its first Config conference in Bengaluru.
The expansion includes integrations involving Amp, ClickUp, Dia, JetBrains’ Junie, Linear, Meta Muse, Mistral, Notion, OpenCode and Sarvam Agents. Figma intends to support multiple interoperability standards while maintaining controls over the information available to external systems.
Field acknowledged growing demand among developers for easier access to design information through AI applications. However, he argued that the existing protocol favours leading artificial intelligence laboratories and does not adequately address differences in how participating companies share information.
Speaking at a media briefing, Field described MCP as an imperfect technology with considerable scope for improvement. He emphasised that Figma wanted customers to use their design information across applications, but said wider interoperability should not automatically require unrestricted disclosure.
The distinction reflects a central challenge confronting software companies as AI agents become increasingly capable of accessing files, retrieving information and performing tasks across platforms. Greater connectivity can improve productivity, but broader access also raises questions about permissions, confidentiality and potential misuse.
Field said exposing additional information through MCP increases opportunities for abuse. Although most customers have legitimate intentions when connecting AI applications to Figma, the company must consider circumstances in which access could be exploited.
His position places Figma between competing demands for open technical standards and tighter control over proprietary information. Developers favour predictable connections between services, while software providers remain responsible for protecting customer data and managing access to their systems.
The disagreement became public after criticism of Figma’s failure to support Pi, an open-source AI coding tool. One of MCP’s creators challenged the company’s approach, prompting Field to question whether the organisations advocating openness were applying comparable standards to their own products.
Field specifically highlighted Anthropic, the AI developer behind MCP, questioning why certain resources created within its ecosystem were not accessible through the same protocol.
Writing on X on September 30, Field argued that discussions about openness overlooked unequal information sharing between technology providers. He pointed to Anthropic’s treatment of artefacts and other cloud resources as an example of the wider problem.
Figma subsequently indicated that it planned to support Pi, signalling that the dispute did not amount to a rejection of the application or the underlying interoperability standard.
The company’s latest partnership expansion also demonstrates that its reservations have not prevented wider adoption of MCP. The protocol provides a common mechanism through which AI applications can interact with external tools and retrieve contextual information needed to complete tasks.
For designers and software developers, such connections can reduce the need to transfer information manually between applications. Design specifications, interface components and related project details can become available within coding environments, potentially simplifying the transition from visual designs to working software.
However, the extent of information available through an integration depends on the capabilities and permissions implemented by the participating services. Supporting a common protocol does not necessarily mean that every application exposes identical resources or permits the same operations.
Figma has been developing additional tools intended to connect design activities with AI-assisted software production. Its October product announcements included expanded agent capabilities, improvements to code-related workflows and new methods for organising project instructions.
The company introduced Specs, a feature designed to accommodate Markdown-based documentation directly within its collaborative design environment. Such documentation can provide instructions and contextual information for developers and AI agents working on software projects.
Figma has also introduced facilities for incorporating design guidelines and code sources into agent-assisted workflows. These features are intended to help automated systems follow established product requirements rather than generate interfaces without sufficient project context.
The developments accompany growing interest in connecting design software with coding agents as businesses experiment with automated application development. Figma is seeking to maintain its role in those workflows while accommodating external AI systems.
Among the newly announced partners, Sarvam Agents adds another participant to Figma’s expanding integration network. The Bengaluru conference also provided a platform for demonstrating how design information can be incorporated into AI-supported development processes across different applications and services.
Follow Arabian Post
Select Arabian Post as your preferred source on Google and MSN News for trusted business news and Arab politics and updates.
