OpenAI Updates Codex for Figma-to-Code Workflows

OpenAI Developers have announced updates to the Codex model that enable a seamless iterative workflow between Figma designs and code. A demonstration video shows the model facilitating a continuous loop from code generation to Figma design implementation and back to code. This integration aims to streamline the process for developers and designers working on user interfaces.

This integration is part of a deepening partnership between OpenAI and Figma; Figma already uses OpenAI models for features on its platform and was an early partner for a ChatGPT app in 2025. The new workflow is enabled by Figma's Model Context Protocol (MCP) server, which allows AI agents like Codex to interface directly with design files. The update comes as Codex is seeing significant growth, with OpenAI reporting over 1 million weekly users and a 400% increase in usage since the beginning of the year. This move also intensifies the competition in AI-powered design-to-code tools, as Figma announced a similar integration with Anthropic's Claude Code just a week prior. This bidirectional, or "roundtrip," workflow allows changes in code to be reflected back in Figma designs and vice-versa, aiming to reduce the friction and manual handoffs that traditionally slow down product development cycles. The goal is to create a more fluid process where engineers can iterate on designs visually and designers can work closer to the final implementation. This advancement surfaces broader discussions on human-AI collaboration, framing AI as a partner that augments creative and technical roles rather than replacing them. The focus is on combining AI's ability to handle repetitive, data-heavy tasks with human creativity, judgment, and oversight to achieve better outcomes. However, the increasing role of AI in creative output raises complex questions about authorship and ownership. Current legal frameworks generally require human authorship for copyright protection, meaning works generated solely by AI may not be protectable. This leads to ongoing debates about who holds the rights when a work is co-created with an AI: the user, the AI developer, or if the work enters the public domain. The ecosystem of AI developer tools is expanding rapidly, with several alternatives to Codex gaining traction. Tools like GitHub Copilot, Cursor (an AI-first code editor), and Windsurf are also offering "agentic" capabilities, where the AI can handle more complex, multi-step tasks. This reflects a broader trend toward more autonomous AI assistants that integrate directly into developer and designer workflows.

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.