
In just over a year, Webflow has gone from being a traditional visual building tool to becoming a platform where AI is woven into almost every workflow. AI Site Builder generates entire websites from one prompt. AI Assistant creates sections, copy, and code components directly in Designer. The MCP server connects Webflow to Claude, Cursor and other AI tools. For Norwegian and Danish agencies and companies, this means that the construction time is significantly shorter than a year ago, while at the same time completely new questions arise regarding quality, control and compliance with European regulations.
There's a lot going on at the same time. Launched in February 2025, AI Site Builder was further developed in February 2026 and can now generate multi-page websites with structure, styles and animations based on a prompt. AI Assistant went from being a contextual aid to becoming a full-fledged collaborator that understands your design system and can create sections that fit in. In April 2026, the Assistant gained the ability to generate production-ready code components directly on canvas. Webflow launched the Claude connector in February 2026, and the MCP server now has access to both Data API and Designer API.
That means AI doesn't just help write copy or generate images. It can actually plan, design, and code in Webflow in a way that previously required manual labor or developer hours.
AI Assistant is at the core of everyday use for most people. It can generate new sections that fit the design system, write copy that matches the brand voice, and build full-fledged web apps with natural language. MCP, or Model Context Protocol, is the more technical but more powerful component. With the Webflow MCP server, developers can connect tools like Claude and Cursor directly to the project and let the AI read and write to the CMS, make changes to the design and pull data out of the structure. For an agency, that means a developer can ask Claude to update all CMS listings of a specific type, or let an AI agent suggest SEO improvements across the entire site without manual clicking.
In our experience, there are some clear areas where Webflow AI adds value in everyday life. Generating sections that fit the existing design system works surprisingly well, saving time on “yet another feature grid or testimonial section”. The first draft of the English copy can be used as a basis for further editing. Code components that previously required developer time can now be generated in minutes for typical applications such as price calculators, booking tools, or interactive calculators. For B2B businesses that launch landing pages quickly or build out content in multiple languages, the speed gains are significant. That doesn't mean that AI will finish building the website on its own, but that the build process can be 30 to 50 percent shorter on many typical tasks.
Here we must be clear. Webflow stores customer data primarily on U.S. servers via AWS, although content is delivered globally via CloudFront. For Norwegian and Danish companies with strict data residency requirements, and especially for the public sector and companies that process sensitive personal data, this is a real consideration. Webflow holds SOC 2 Type II, ISO 27001, and several other certifications, and states that they are actively working towards ISO 42001 and compliance with the EU AI Act, but are not fully certified against these frameworks.
The EU AI Act will be fully applicable from 2 August 2026. Norwegian companies are required through the EEA to comply with the regulations. There are proposals for a postponement until December 2027 via Digital Omnibus, but the practice is to reschedule with the original date. For the use of AI as pure productivity enhancement in building web pages, this is basically low-risk, but if you use AI for functions that affect users' rights, or process sensitive data, classification needs to be considered concretely.
Webflow AI builds on generative language models that are primarily trained in English. AI generated copy in Norwegian and Danish works surprisingly well for simple formulations, but often needs more manual tweaking than in English, especially for subject language, B2B tone and industry-specific terminology. AI can give you a first draft that gets 60 to 70 percent of the way, but the final 30 to 40 percent that actually makes the text marketable requires a human editor with linguistic flair. Multilingual setup in Webflow Localization works technically, but structure, hreflang tags and local SEO for Norwegian and Danish are still manual work. AI Assistant is not very smart in this area yet.
Before you go in with Webflow AI as part of the build process, a few things should be clarified:
The honest truth is that AI is fantastic at the repetitive, but still weak at the strategic. We at WeAssist use Webflow AI every day for customers. We use AI Assistant to build out sections in an existing template, MCP the server to major changes to the CMS, and code generation for specific interactive elements. But decisions about information architecture, conversion strategy, brand expression, complex multilingual setup and compliance with GDPR and the EU AI Act are still human judgments. AI does 30 to 50 percent of the work faster. The last, most important part is still ours.
Wondering how Webflow AI can be used for your business and where you should be careful? We help Norwegian and Nordic companies set up Webflow projects that leverage AI where it adds value, while taking into account privacy, language and design controls. Get in touch, then we find out how much AI can help your particular project, and where you still need human expertise.