AI can generate interfaces quickly. Distinctive products still need context and judgment.
AI has accelerated the familiar template problem.
Generative tools can produce polished interfaces quickly. Without a real brand, audience and product context, the output can feel interchangeable: the same cards, softened gradients, oversized type and generic “AI assistant” panel.
Technically, the output is competent.
This is not new. Template marketplaces created the same problem years ago. When everybody buys from the same visual library, differentiation moves elsewhere.
AI simply accelerates the effect.
The solution is not to reject generative tools.
Figma’s move toward structured MCP access and design-system-aware agents is significant because it shifts AI away from generic visual prediction and toward organisation-specific systems.
An agent that knows your real components, tokens, typography, interaction rules and production constraints is much more useful than one that knows only what a modern SaaS dashboard usually looks like.
This has consequences for agencies.
Clients can buy screens in many ways. An agency earns its fee by making the experience coherent with the organisation, its audience and its operating model.
If AI makes the commodity layer cheaper, then agencies need to stop competing on commodity output.
The valuable work becomes:
- defining the design system
- setting interaction principles
- establishing information architecture
- connecting real data and workflows
- making automation understandable
- maintaining accessibility
- testing edge cases
- aligning the product with brand and business intent
AI can accelerate each of these areas, but it cannot replace the need for a point of view.
The danger is that agencies will use AI to produce more of the same things they were already producing, only faster.
That creates efficiency without differentiation.
The opportunity is the opposite.
Use AI to remove repetitive production so more attention can go into the decisions that actually make a product distinct.


