Artificial intelligence’s first act was conversational. Models could write emails, generate marketing copy, and sketch prototypes. But the consumer fascination with prompts and chat interfaces hid a harder problem: how to make AI dependable inside enterprise systems that run on compliance, security, and change control.
There are now signs of progress. ServiceNow and Figma announced a technical collaboration that allows a design file to become a working enterprise application. Through a new Model Context Protocol (MCP) integration, ServiceNow’s AI Build Agent can ingest a Figma design link, interpret every component and variable, and generate a deployable workflow inside the ServiceNow platform. Early pilots report that initial user-interface and data-model implementation time fell by more than 80 percent.
“Great product design makes a positive human impact when it becomes a real experience in the hands of people,” said Amy Lokey, ServiceNow’s chief experience officer. “By connecting Figma with ServiceNow’s AI-powered workflows, we’re making it effortless for teams to turn ideas into enterprise-ready applications in minutes.”
The design-to-app pipeline arrives
Until now, most design-to-code experiments ended at prototypes. The problem wasn’t creativity; it was semantics. Design tools captured colors and shapes, not logic. MCP changes that by providing a common, JSON-RPC-based protocol for exchanging structured context between AI agents and external systems.
Figma’s own MCP server exposes design metadata (component names, style tokens, variables) so a consuming agent can treat a file as a structured model, not a flat image.
Inside ServiceNow’s Zurich release, the Build Agent parses that semantic layer, maps it to UI elements, and scaffolds data models, logic, and tests in a sandboxed development environment. The company calls the workflow “vibe coding”: conversational prompts produce real application assets that still pass through the platform’s governance stack (OAuth 2.0 authentication, audit logs, version control, and AI Control Tower policies.)
“In a world of AI-generated software, design is the differentiator that will make your product stand out,” said Kris Rasmussen, Figma’s CTO. “This MCP integration brings important Figma design context directly into ServiceNow’s AI workflows to help teams efficiently build high-quality, differentiated enterprise products.”
ServiceNow’s 900-person design organization already runs its design system in Figma, allowing designers, engineers, and product managers to collaborate in the same files. The new pipeline extends that workflow downstream: from shared design to governed deployment.
Market data suggests the timing is right. Grand View Research values the global low-code application platform market at US$24.8 billion in 2023 and projects it will reach US$101.68 billion by 2030, a compound annual growth rate of 22.5 percent.
Mendix reports that 98 percent of enterprises already use low-code tools, 85 percent say combining AI and low-code helps them innovate faster, and 71 percent cite governance as their top concern.
Gartner defines enterprise low-code platforms as offering “model-driven development tools, generative AI, and prebuilt component catalogs… collaborative development… runtime environments… governance controls… agentic AI and integration capabilities.”
The Figma-ServiceNow integration targets that gap directly: speed without loss of control.
The new AI stack is about integration, not prompts
Bill McDermott, ServiceNow’s CEO, has spent the year reframing enterprise AI as a systems problem rather than a language one. “AI is a cross-functional sport,” he told investors in October. “It’s the biggest breakthrough in enterprise technology in half a century.”
He argues that AI’s value won’t come from large-language-model providers alone but from workflow platforms that can govern their output. “Those agents are being sold into silos, and that’s the very reason why AI won’t work.”
That perspective positions the Figma partnership as more than a feature. It’s a template for how generative design, code, and governance can coexist. ServiceNow’s Zurich platform added ReleaseOps, Vault, and Agentic Playbooks to control how autonomous agents behave during development and deployment. Figma’s MCP server uses the same principles of containment and traceability: server-to-server authentication and token storage within the customer instance.
The move also helps Figma prove its post-IPO direction. After crossing a US$1 billion annual run rate and reporting 38 percent year-over-year growth in Q3 2025, the company is under pressure to translate its AI investment into enterprise revenue.
The ServiceNow collaboration gives it a credible story beyond creative teams: one rooted in integration velocity, not design flair.
Analysts see that pivot across the sector. Forrester and IDC highlight cross-functional governance as the next bottleneck in AI deployment, while Gartner’s 2025 market definitions formalize “agentic AI” as a core capability for enterprise platforms.
In that context, ServiceNow and Figma are offering a playbook for AI’s industrial phase that treats creative input as structured data and subjects AI output to the same rigor as code.
From interface to infrastructure
Where the first wave of AI tools sold ease of use, the next wave sells accountability. Enterprises don’t want AI that writes; they want AI that builds under governance. The MCP protocol and ServiceNow’s agentic framework together mark a standardization of that goal.
There are still risks. MCP introduces new attack surfaces, and third-party vulnerabilities have already required patches for command-injection exploits.
But the direction is clear: AI that remains sandboxed is safe AI, and AI that integrates is valuable AI. According to McDermott, enterprises “can’t duplicate what [ServiceNow] took 20 years to build.”
