Webgate used Elementor MCP with Claude Code and Codex to build a landing page on its live agency website, webgate.digital. The agency kept its existing Elementor V3 components and added new V4 Atomic Elements, testing an AI-assisted workflow on a site already in use rather than a blank demo.
For agencies weighing how to add AI tools to an established WordPress project, the case offers a practical example: provide the agent with the brief and design context, let it build native Elementor structures, then review the result in the browser and editor. The team kept people responsible for design decisions and final publishing.
In brief: Webgate’s workflow moved from a written brief and existing site references to prompts, Elementor MCP, visual QA and refinement. Claude Code and Codex helped create and revise page sections, while the team mixed new Atomic Elements with existing V3 modules. It did not rebuild the five-year-old website or replace its established setup.
How Webgate used Elementor MCP on a production site
Webgate co-founder Tomáš Jendek described the project in a case study published by Elementor on September 23, 2026. The work focused on a landing page for the agency’s AI-native website development approach, built on its existing production WordPress site.
The site already had years of Elementor V3 components, global styles and content. Webgate wanted to find out whether it could introduce agent-assisted development without rebuilding the site or disrupting that V3 setup. Its working stack included WordPress, Elementor Pro, Claude Code, Codex, Figma references and browser-based visual checks.
The process was straightforward: brief, existing website context, screenshots or Figma, prompts, Elementor MCP, visual QA and refinement. Some page sections reused existing V3 structures. Others were created as new V4 Atomic Elements through the connection.
Elementor MCP connects agents to native page structures
Elementor introduced its MCP connection on September 16, 2026, according to its launch announcement. MCP, or Model Context Protocol, gives compatible AI tools a way to interact with Elementor. Instead of returning only a mockup or instructions to copy by hand, an agent can work with Elementor structures directly.
That distinction matters in an agency workflow. The output remains editable in Elementor, so a developer can inspect the page, adjust it in the visual editor and continue working with the same site components. Elementor says MCP-generated work is saved as a draft rather than published automatically.
Where the agents helped
Webgate used the tools for specific implementation tasks, rather than asking an agent to create an entire site from a broad instruction. The team described building sections from written briefs, working from screenshots, translating Figma designs into native Elementor structures, reusing existing V3 modules and iterating on layouts with prompts.
Agents also helped with combined content and structure updates. With the page context available, a developer could ask for copy changes alongside layout changes instead of treating those as separate rounds of manual editing. The case study describes this as a way to reduce repetitive assembly work, not as a measured time saving.
Keep the working design system in context
The source material for a prompt was part of the workflow. A written brief could describe the intended section; screenshots could communicate visual direction; and Figma frames could guide the translation into Elementor. Existing site patterns and design variables gave the agent context for how new work should fit the page.
That context also let Webgate choose different approaches within a single page. Reusing a working V3 module made sense where it already fit. New Atomic structures were used for sections that needed to be built differently. The project did not require an all-at-once migration from V3.
Visual QA stayed in the human workflow
Generating the initial structure was only one part of the process. Webgate checked the rendered page in a browser, compared it with the reference, and sent further instructions when layout or spacing needed work. The team could then make structural changes through the agent or handle small visual adjustments directly in Elementor.
That division of work is sensible for production sites. A prompt can specify a layout change, but people still need to judge whether the information architecture makes sense, the visual hierarchy is clear and the page works across screen sizes. Webgate kept human review in charge of UX, accessibility, responsive behavior, design consistency and final publishing.
The case is useful because it addresses a common agency constraint: established sites cannot always be rebuilt just to adopt a new development method. Webgate added AI-assisted work to a five-year-old site while keeping its existing V3 setup in place. The reported result is a workflow model, not a published performance benchmark or quantified productivity gain.
What agencies can take from the case
Webgate’s example points to a gradual approach. Start with a defined page or section, provide the relevant site and design references, and ask the agent for a specific Elementor implementation. Review the rendered result, make corrections, and retain the existing components that still serve the project.
That leaves the agent doing much of the mechanical assembly while developers set direction and check the work. It also avoids treating AI adoption as a reason to replace an established site architecture. For agencies with active client sites, that is a more realistic starting point than an unrestricted redesign.
FAQ
What did Webgate build with Elementor MCP?
Webgate built a landing page on its live agency website, webgate.digital. It combined existing Elementor V3 modules with new V4 Atomic Elements.
Did Webgate rebuild its site in Elementor V4?
No. The agency added new Atomic sections while keeping its existing V3 setup and components. The case study describes a gradual addition, not a full-site migration.
Did AI publish the page without review?
No. Webgate kept people responsible for visual QA, accessibility, responsive behavior, design consistency and final publishing. Elementor’s launch information also says MCP-generated work is saved as a draft rather than published automatically.
Agencies considering a similar workflow should start with a low-risk landing page or section, supply the agent with current design references, and review the draft in both the browser and Elementor editor before publishing.
Editor’s note, November 17, 2026: For a related workflow outside WordPress, Spacefast can publish an AI-built prototype as a separate Space. It doesn’t create native Elementor components or replace the site’s editor. See our step-by-step Spacefast guide for the steps. This is an affiliate link; WP Gazette may earn a commission from qualifying purchases at no extra cost to you.