Your AI tool is only as smart as the context you give it.

The numbers don't lie. According to Figma's State of the Designer 2026 report, based on a survey of 906 digital designers across North America, Europe, APAC, LATAM and the Middle East, 72% are now using generative AI in their daily workflows. Of those, 91% say it improves the quality of their work. 89% say it makes them faster. 

So why does AI still feel clunky for so many teams? 

Simple. Most people are still using it like a search engine. Ask a question. Get an answer. Move on. But that's just scratching the surface of what's possible when AI actually understands your world. 

Enter the MCP. 

A Model Context Protocol (MCP) is a structured, secure way of connecting an AI tool to the systems your team already uses. Think of it less like giving AI a login, and more like writing it a proper onboarding document. 

When a new team member joins, you don't expect them to figure everything out from scratch. You give them context: here's how we work, here's what you can access, here's what's off limits. An MCP does exactly that for AI. It tells the tool what it can see, what it can do, and crucially, what it can't touch. 

The result? An AI that works with your systems, not around them. 

What that looks like in practice. 

We've been trialling this on our own sites, and two things stand out immediately. 

Content awareness from day one. Instead of spending hours briefing an AI on your site structure, content model, or naming conventions, it can explore and understand those things for itself, because the MCP gives it direct, controlled access. Less briefing. Fewer errors. Better outputs. 

Dramatically less manual grind. Picture a content team that needs to publish dozens of new pages across a large site. Normally that's hours of copying from a spreadsheet into a CMS, field by field, page by page. With an MCP-connected AI, that entire process is restructured. The AI agents read structured content, understands the system it's working in, and handles the publishing at scale, with humans staying in control throughout. A full day's work, done before lunch. 

A note on access. 

It's built around controlled, permissioned access. Read content? Yes. Browse the schema? Sure. Delete the database? Absolutely not. The AI operates in a defined lane and nothing more. As we develop our thinking here, that principle of human oversight stays non-negotiable. 

Why this matters right now. 

The Figma report makes one thing very clear: the teams seeing the real benefit from AI aren't the ones experimenting occasionally. They're the ones who've embedded it properly. Designers whose AI use is growing are 25% more likely to report increased job satisfaction. Their companies are more likely to be growing faster than the industry average. 

That shift, from occasional tool to integrated team member, is exactly what MCP enables. Not replacing human judgement. Not cutting corners on craft. Just removing the friction that gets in the way of good work. 

We believe technology should remove barriers, not create them. MCP does exactly that. It strips away the friction between people and their tools, and lets teams focus on the work that actually matters.