Key takeaways
- AI generates the average of everything it's seen — UX context makes it produce the right thing for your users and brand.
- The most important design decisions happen in error and edge states, exactly where AI is weakest.
- A mature design system is the guardrail that keeps AI-generated output coherent and on-brand.
- Use AI for breadth and speed; keep humans in charge of decisions, research, and failure states.
- Context, not screen volume, is the real competitive advantage in AI-assisted design.
AI can now produce a working interface faster than a team can finish its kickoff meeting. That speed is seductive, and it is reshaping how design gets done. But there is a quiet failure mode hiding in the excitement: models generate the average of everything they have seen, not the right thing for your users, your brand, and your business. The gap between those two is exactly where UX knowledge lives. We call the discipline of closing that gap UX-context design.
Why raw AI output isn't design
Generation is not the same as decision-making. A model can lay out a checkout flow, but it cannot tell you why your customers abandon at step three, which stakeholder assumption is quietly wrong, or how the interface should behave when everything breaks. Good design has always been less about producing screens and more about resolving tension — between user needs, brand strategy, technical reality, and commercial goals.
The UX Collective's "Kintsugi problem" captures this beautifully: the decisions that matter most are made when things go wrong, not when the happy path works. AI is excellent at the happy path. It is largely blind to the empty states, the error recovery, the edge cases where trust is won or lost. If you accept the machine's first draft, you inherit a product that looks finished and behaves fragile.
Context is the real prompt
The strongest visual and interaction concepts never start in a tool — they start with the right questions. Before a single component is generated, senior teams do the unglamorous work: interrogating brand strategy, surfacing hidden stakeholder assumptions, and agreeing on a shared direction. This pre-concept phase is what turns a vague brief into a defensible foundation.
When you feed that foundation into an AI workflow, the output changes character. Instead of a generic dashboard, you get one shaped by known constraints. UX-context design is essentially the practice of front-loading this knowledge so the machine has something meaningful to reason against.
- Brand context: tone, positioning, and the emotional register the interface should carry.
- User context: real behaviours, pain points, and the moments where the journey is fragile.
- System context: the components, tokens, and patterns your product already relies on.
- Failure context: what happens on empty, error, slow, and offline states.
Your design system is the guardrail
An AI that generates against a mature design system produces coherent output. An AI that generates against nothing produces a museum of inconsistent screens. Nielsen Norman Group's six-dimension maturity model is a useful lens here: the healthier your system's adoption, governance, and documentation, the more you can safely automate. A weak system amplified by AI simply scales its incoherence faster.
Practically, this means treating your design system as the primary context you give any generative tool. Well-named tokens, documented usage rules, and clear component logic act as guardrails that keep machine output on-brand and on-pattern. The system becomes the difference between AI that accelerates you and AI that quietly erodes your product's quality.
Speed without losing judgement
When 32 design leaders were asked what they do when told to move faster, the wiser answer echoed something the Greeks understood: festina lente — make haste slowly. Speed is worthless if you accelerate in the wrong direction. AI hands you enormous velocity; UX context supplies the steering.
There is also a deeper risk worth naming. As design increasingly shapes the whole organisation, and machines increasingly shape design, we can drift into pure efficiency — products that decide where you'll end up before you've had a chance to explore. Great digital experiences still leave room to wander, to be surprised, to play. That intentional friction is a human judgement call. No model will make it for you.
How we apply UX-context design at Evocative
Our approach is to keep AI in the loop but never in charge. We use generative tools to explore breadth quickly — many directions, many states, many variations. Then human UX judgement does the narrowing: validating against research, stress-testing failure states, and enforcing design-system integrity.
- Start with strategy and research, not the tool.
- Encode brand and system context before generating anything.
- Use AI for breadth and speed; use people for decisions and edge cases.
- Measure the output against design-system maturity, not just visual polish.
The teams that win the next few years won't be the ones generating the most screens. They'll be the ones with the richest context — and the discipline to apply it. That is where business value actually accrues: fewer redesigns, faster shipping, and interfaces that hold up when reality gets messy.
Sources
- Design-System Maturity: A 6-Dimension Framework — Nielsen Norman Group
- From Kickoff To First Concept: How To Turn Brand Strategy Into Visual Direction — Smashing Magazine
- The Kintsugi problem — UX Collective
- When design becomes the organisation, and AI becomes the design — UX Collective
- Play and design: in favour of the novel digital experience — UX Collective
- What 32 design leaders do when told to move faster — UX Collective
