c-suite

Most design systems are not ready for AI

insight
June 23, 2026
9 min read

Author

Noel Cunningham

Managing Director and Head of Nagarro's Global Design Studio. A human-centered design leader, she specializes in experience strategy, design transformation, and innovation, helping organizations create impactful products and experiences that drive business value.

Every AI tool a business uses can have an outsize impact on its operations and success. Perhaps that is most important when it comes to using AI for design, UX and marketing. In other words, how your brand is viewed can be heavily dependent on the choices you make, using AI. Companies are spending enormous amounts of time debating which models and assistants to adopt. Far less attention, unfortunately, is being paid to which systems those tools depend on to produce consistent work. 

Consider this example: A developer opens a chat interface and types, "Build a settings page." Thirty seconds later, there's a working component. The code runs, the layout holds together, and what was previously a time-consuming task has been efficiently handled. But then consider what happens when a user looks closer. The font and spacing are slightly off.. The result looks like every other product on the internet. The page never gets flagged in review because technically, nothing is wrong, and eventually it ships. 

The end user, ultimately, looks elsewhere and finds a brand that presents itself more uniquely, intuitively, and impressively. 

Now scale that challenge across dozens of developers, multiple AI tools, and hundreds of product decisions over the course of a quarter. Small inconsistencies stop being isolated issues and start shaping the product experience and impacting the business's overall success.

The brief nobody wrote

Most of the current conversation is about which AI tool to adopt: Cursor or Copilot? v0 or Bolt? Claude or GPT? The debate is real, and the tools are impressive. But it's focused on the wrong problem. 

Many organizations are introducing AI into workflows that were never structured clearly enough in the first place. For a long time, design systems worked because they didn’t need to carry everything explicitly. They were supported by human context. Designers understood intent without it being fully written down, engineers learned patterns through repetition, and teams relied on shared judgment built over time. 

AI does not work that way. It doesn’t infer intent from culture or context. It only responds to what is explicitly defined. 

That difference is now visible in day-to-day production work. A system might define components, colors, and typography, but still leave open how and when those elements should be used. Those decisions often live outside the system itself, in conversations, habits, and institutional memory. That gap is where inconsistency begins to build. 

The silent failure nobody catches

There is a failure mode in AI-assisted work that isn’t obvious. It looks like work that passes review, code that runs, and copy that sounds acceptable. The issues are subtle enough to pass through traditional review cycles.  

This is where the gap shows up - between tools that can produce output and systems that can govern it. Without governance, AI can produce work that looks correct at first glance, but doesn’t match the system behind it.  Over time, that’s where the drift starts to appear. 

A button that's almost right. A color that's close. A tone that's slightly off. None of these are significant on their own, and they don’t trigger alarms or block releases. But over time, they compound across features, teams, and release cycles. The product starts to lose consistency in ways that are hard to trace back to a single decision. 

Every quarter, more of the product is touched by AI tools. Features are prototyped, copy is drafted, and interfaces are generated. Each one is a moment where output either aligns with your brand or moves away from it. The speed that makes AI so valuable is the same speed that makes it risky when there's nothing guiding it. 

Speed becomes a liability when there's nothing for AI to be loyal to, i.e., the absence of guiding principles.

What "AI-Ready" actually means

AI-readiness is not a feature you add. It's about clarity. If a new system started working with your design infrastructure today with no human augmentation, what would it actually be able to understand? 

In most organizations, the honest answer is: not much. AI may recognize components and tokens, but it won’t understand intent, tradeoffs, or the reasoning behind key decisions. It won’t know why one pattern is preferred over another, or when a technically valid option should not be used. 

Those brand distinctions matter more in an AI-assisted workflow because the volume of generated output increases dramatically. Without clear guidance, the system produces plausible results that slowly move away from the intended experience. A model can identify what exists in a design system, but it cannot reliably determine how it should be applied unless that logic is part of the system itself. That’s where most teams are starting to feel friction - keeping outputs consistent with the system. 

"AI can scale creation at unprecedented speed, but it cannot protect what it doesn't understand. Without a clear source of truth, every output becomes a step away from your identity. Speed becomes a liability when there's nothing for AI to be loyal to."

-Noel Cunningham

The real shift

What’s changing is who is contributing to the work in the first place. For most of design history, the designer was the primary author of the product. They made the decisions in the file, the review, and during handoff. AI doesn't remove the designer from that process, but it does introduce a new author between intent and output. That author is fast, tireless, and completely indifferent to a brand unless you've made your brand impossible to ignore. 

Without structure, AI pulls from whatever context it can find and generates something that seems reasonable. When the system is clearly defined, the output becomes more consistent because the constraints are clearer. 

That compounds over time. Teams spend less time revisiting decisions, correcting inconsistencies, or rebuilding work that drifted off course early in the process. The design system stops functioning as reference material and starts shaping how products actually get built. 

Teams moving quickly with AI are already running into this problem. The outputs improve when the underlying system is easier to interpret and apply consistently. 

A new mandate

Design systems were originally built for people already close to the work - designers, engineers, and product teams who understood the context around the product. Much of that context never had to be written down because teams carried it through conversations, reviews, and experience. That is starting to change. 

AI tools are now participating directly in product development. They generate interfaces, write copy, suggest code, and make decisions that affect how products look and behave. The more these tools are used, the more pressure there is on the underlying system to provide clear guidance. 

In many companies, important decisions still live outside the system itself. Teams know which patterns feel off-brand. They know when to avoid a component even if it technically works. They know the difference between something that is functional and something that actually feels right for the product. AI does not automatically understand any of that context. 

The companies handling this well are building systems that are easier to interpret, easier to apply consistently, and less dependent on institutional knowledge that only exists inside the team. 

The underlying system starts to matter much more once AI becomes embedded in everyday product development. Teams can move quickly with AI, but keeping the product consistent depends on how clearly the system defines what good output actually looks like. 

The companies that solve this early will have an advantage that goes well beyond speed: they’ll be able to scale AI-generated work without losing the identity of the product itself. 

Building AI-ready products requires more than powerful models. It requires design systems built for intelligence, adaptability, and trust.

Is your design system AI-ready?

Get in touch