Figma is moving to formalize how artificial intelligence agents behave inside its product suite, a notable step in the company's broader effort to turn design software into a more programmable, collaborative workspace. The new features are designed to define the scope, permissions, and working patterns of agents on the platform, reflecting a growing industry shift from generic AI helpers toward controlled, task-specific systems.
The announcement comes as Figma also releases its code layers feature in open beta, giving users broader access to a tool intended to connect visual design work with underlying code structures. Together, the two updates suggest Figma is not simply adding AI for novelty, but building a framework in which designers, developers, and automated agents can work in closer alignment.
Agentic Workflows
Figma's emphasis on agent behavior is significant because the market is rapidly moving beyond chat-style assistants and toward agents that can take actions across software environments. In practical terms, that means companies are increasingly asking not just what an AI can say, but what it is allowed to do, how it should be constrained, and how its output can be trusted inside production workflows.
For a platform used by product teams, those questions matter. Design systems often involve multiple contributors, version control concerns, and handoffs between design and engineering. By introducing features that define how agents work, Figma is positioning itself as a governance layer for AI-assisted creation rather than a passive host for third-party tools.
That approach may prove especially important as enterprises look for predictable controls around automation. The more an agent can interact with files, components, and code-adjacent assets, the more essential it becomes to set boundaries. Figma's move appears aimed at making those boundaries explicit, which could help reduce friction for teams that want AI assistance without surrendering oversight.
Code Meets Design
The open beta for code layers is equally strategic. Figma has long been central to visual collaboration, but the company has increasingly sought to reduce the distance between design intent and implementation. Code layers are part of that effort, offering a way to represent code-linked elements more directly within the design environment.
By opening the feature to beta users, Figma is broadening testing while also inviting developers and designers to evaluate how well the tool fits real-world workflows. The timing matters: as AI coding tools proliferate, design platforms are under pressure to become more interoperable with engineering systems. Figma's latest release suggests it wants to remain the place where those workflows begin, not merely where they are handed off.
The combination of code layers and agent controls also points to a deeper product thesis. If AI agents are going to participate in design and development work, they need structured access to both visual and technical layers of a project. Figma appears to be building toward that integrated model, where automation is not bolted on but embedded in the workflow architecture itself.
Competitive Stakes
The broader competitive context is clear. Frontier AI companies and productivity platforms are racing to define the next interface between human users and machine assistance. Some are focused on general-purpose agents; others are embedding AI into specialized vertical tools. Figma's strategy sits in the middle: it is using domain depth in design collaboration to create a more controlled environment for agentic work.
That could be a meaningful differentiator. In enterprise software, trust often matters as much as capability. A feature that is powerful but poorly bounded can create operational risk, while a feature that is constrained but reliable may be easier to adopt at scale. Figma's latest announcements indicate an awareness of that tradeoff.
The company's challenge now will be execution. Users will judge whether these features genuinely improve speed, clarity, and collaboration, or whether they add complexity to already intricate workflows. But the direction is unmistakable: Figma is betting that the future of creative software will be shaped not only by AI output, but by the rules that govern AI action.