Figma’s latest update moves the product closer to a shared working surface for designers and engineers rather than a handoff tool between them. The headline change is a new code layer on the collaborative canvas, which can clone repositories and map code back to design layers for testing. In practice, that means teams can inspect implementation details alongside the UI they are building, tightening feedback loops when a component’s behavior diverges from the spec.
That shift matters because it changes where iteration happens. Instead of treating design files and codebases as separate systems reconciled late in the process, Figma is now trying to keep the conversation inside the canvas. The company has already spent the past year moving in that direction with Figma Make and integrations with Claude Code and Codex, but code layers go a step further by bringing repository-level context directly into the design environment. For teams that struggle with translation errors between mockups and implementation, the appeal is obvious: easier comparison, faster debugging, and less back-and-forth over whether a mismatch lives in the design or the code.
Yuhki Yamashita, Figma’s chief product officer, framed the feature as a way to optimize for iteration rather than production-grade code quality. That distinction is important. The code layer workflow is not positioned as a replacement for engineering repositories or build systems; it is a collaborative interface for evaluating flows, testing mapped layers, and iterating on ideas with designers, product managers, and programmers in the same workspace. If teams adopt it that way, the likely benefit is cleaner handoffs and more rapid convergence on UI behavior. If they treat it as a shortcut around disciplined engineering review, the result could be the opposite.
The update also widens what can be expressed directly in Figma. Native support for animations, transitions, and 3D transforms brings motion design closer to the core canvas, which should reduce reliance on external tooling for prototyping interactive states. That can help teams test timing, sequencing, and visual continuity earlier in the process, before those decisions harden into implementation work. At the same time, these additions increase the technical surface area of the design system itself. Motion is harder to standardize than static layout, and 3D transforms can introduce complexity that is easy to showcase and harder to maintain once it ships.
Figma is also adding AI-generated shader effects and fills, alongside a broader push toward AI-assisted plugins for specific tasks. On the creative side, this expands the range of effects designers can explore without dropping into lower-level graphics tooling. On the operational side, it introduces another layer of review. Shader generation and AI-created fills may accelerate experimentation, but teams still need to understand how those assets are produced, whether they can be reproduced reliably, and how they fit into approved design and brand systems. The more the canvas becomes a place where generated visual treatments and code-backed layers coexist, the more important it becomes to track provenance and maintain consistency.
That governance problem becomes more acute when repositories are cloned into a design workflow. Pulling live code into Figma can improve realism during testing, but it also creates new questions about access control, repository scope, and data exposure. Teams will need clear rules about which repositories can be mirrored, who can see them, what parts of a codebase are safe to surface in a collaborative environment, and how changes are synchronized back to engineering. In companies with regulated data, proprietary logic, or tightly controlled environments, those decisions cannot be left to individual teams to improvise.
The same is true for AI integrations. Figma’s Claude Code and Codex connections already point toward a more blended workflow, but once code generation, code inspection, motion design, and shader creation all live in the same product, governance has to extend beyond file permissions. Organizations will need to decide where AI can be used, what outputs require review, and how to prevent generated assets or imported code from bypassing internal standards. The value proposition is workflow compression; the cost is that more of the product lifecycle now sits inside one system that has to be governed like several.
The timing suggests Figma is betting that teams are ready for that tradeoff. Its June 24 rollout announcement lands in a market where design tools increasingly compete on how well they connect to development, not just how polished the canvas feels. If the new workflow holds up in real use, Figma could strengthen its position as a cross-domain collaboration layer spanning design, code, and AI-assisted creation. But success will depend less on feature density than on operational maturity: access policy, repo hygiene, review paths, and clear boundaries between exploratory work and production engineering.
For teams evaluating the update, the right question is not whether code layers and motion tools are impressive. It is whether they can be introduced without weakening the controls that keep software maintainable. Figma is making the canvas more powerful. The challenge for users is making it governable.
