Design platform Figma is significantly enhancing its AI capabilities through strategic partnerships with Anthropic and OpenAI — and the implications for how design and engineering teams collaborate are more significant than the headlines suggest.
What Figma Is Building
AI-Powered Diagram Generation in FigJam
Figma has integrated Anthropic's Claude AI into its FigJam whiteboarding tool, allowing teams to generate editable diagrams from written prompts, PDFs, images, or screenshots. The integration moves AI conversations out of siloed chat interfaces and into collaborative visual workspaces — creating shared artefacts that design, product, and engineering teams can work from together.
The most significant shift here is not the capability itself — it is the workflow change. AI-generated diagrams that live in FigJam rather than a chat window can be reviewed, edited, and acted on by the whole team. That changes how decisions get made.
Bridging Code and Design with OpenAI
In collaboration with OpenAI, Figma is launching a "Code to Canvas" feature that converts code generated by AI tools — including OpenAI Codex — into fully editable designs within Figma. This creates a direct link between AI coding environments and Figma's design process, allowing teams to refine AI-generated interfaces, compare design options, and align on decisions more efficiently.
What This Means in Practice
For Design Teams
The shift toward code-to-canvas and AI diagram generation compresses the feedback loop between design and development. Prototypes can be generated faster. Structural decisions that previously required back-and-forth between tools can now happen within a single platform. For teams with established design systems, the question becomes how AI-generated output is evaluated against system standards — and who owns that review.
For Mission-Driven Organisations
For smaller organisations without dedicated design teams, these integrations lower the barrier to structured visual thinking. A programme director can generate a service blueprint in FigJam without needing design support. A communications lead can create a wireframe-level layout without touching a design tool directly.
The risk is quality control. AI-generated design outputs require review by someone with the judgement to assess whether they are accessible, on-brand, and structurally sound. That review function does not go away — it becomes more important as generation becomes easier.
