Co-design and AI: why humans remain indispensable

What is co-design in the AI era?

Co-design is a design approach that actively involves end users, stakeholders and experts in the process of creating a product or service. In the era of generative AI, this approach is sometimes seen as obsolete: why organise costly workshops when AI can generate mock-ups in seconds?

The answer is straightforward: AI produces artefacts, not understanding. Without a deep grasp of needs, even the prettiest prototype is useless.

What AI can do in design

AI's capabilities in design are real in 2026, driven by the creative trends reshaping the sector:

• Wireframe generation: from a text description, tools such as Galileo AI or Figma AI produce functional mock-ups in seconds.
• Rapid prototyping: AI generates interactive interfaces you can test immediately.
• Design variations: instead of creating 3 variations manually for an A/B test, AI produces 20 in a minute.
• Benchmark analysis: AI analyses hundreds of competitor interfaces to identify dominant patterns.

For French-speaking Swiss SMEs with limited budgets, these capabilities are a valuable accelerator. An entrepreneur can move from idea to testable prototype in a day instead of three weeks.

Measurable gains for design teams

The figures confirm AI's contribution to design workflows. According to a McKinsey study (2025), teams that integrate AI into their design processes reduce prototyping time by 40 to 60%. Forrester estimates that the average cost of a design-sprint cycle drops from around CHF 25'000.– to roughly CHF 12'000.– when AI takes over variation generation and benchmark analysis.

That said, these productivity gains apply only to the mechanical phases of the process. User research, workshop facilitation and result interpretation remain high-value human activities.

What AI cannot do

Empathy and active listening

Co-design begins with listening. An experienced UX designer picks up on hesitations, contradictions and unspoken emotions from a user. They sense that the client says "it's good" while frowning, and dig into that dissonance.

AI perceives none of this. It works from textual or structured data, not from body language, emotional context or human intuition.

Swiss cultural context

The French-speaking Swiss market has specifics that AI struggles to grasp. This is one of the key skills that make a difference in the AI era in Switzerland: mastering local context. The relationship to discretion, the value placed on quality rather than volume, the sensitivity to multilingualism (French, German, English), the high compliance expectations: these cultural nuances shape local users' needs.

A co-design workshop with Zurich and Geneva clients reveals subtle but decisive differences in UX expectations. AI, trained on predominantly Anglo-Saxon data, produces standardised interfaces that ignore these nuances.

Unspoken needs

The most important needs are often those users do not articulate, either because they consider them obvious or because they do not know a solution exists. Co-design surfaces these latent needs through observation, creative exercises and confrontation with prototypes.

AI cannot discover what it has never seen in its training data. It refines the existing; it does not reinvent the experience.

Concrete methodologies: AI-augmented Design Thinking

Design Thinking remains the reference framework for co-design. AI does not replace any of its five stages, but it accelerates them in targeted ways.

Empathy: collect more, analyse faster

Before a workshop, AI can analyse thousands of customer reviews, support tickets and social-media verbatims. It produces a thematic mapping of frustrations and expectations in a few hours, where a human analyst would take several days. The designer then enters the workshop with a solid factual base, ready to validate or invalidate these hypotheses through direct observation.

Ideation: multiplying creative paths

In the ideation phase, AI generates dozens of visual concepts from the collected insights. The team no longer starts from a blank page: it selects, combines and adapts AI's proposals. This approach reduces the conformity bias that often limits classic brainstorming.

Prototyping and testing: iterating in real time

During user-testing sessions, a designer can modify a prototype live thanks to AI. A participant expresses a preference for different navigation? The designer generates a variation in less than two minutes and immediately submits it to the group. This responsiveness turns testing sessions into genuine co-creation workshops.

Practical tools for Swiss teams

Several tools allow you to integrate AI into a co-design process without heavy investment:

• Figma AI (Figma): mock-up generation, smart auto-layout, component suggestions. Natively built into the tool that most design teams already use.
• Maze: a user-testing platform that integrates automated journey and friction analysis. Ideal for remote testing with users spread across Geneva, Lausanne and Zurich.
• Miro AI: assistant integrated into collaborative whiteboards to structure workshop outputs, group sticky notes by theme and generate summaries.
• Dovetail: AI-assisted qualitative analysis of user interviews. Transcription, thematic coding and insight extraction in a few clicks.

The common thread of these tools: they augment analytical capacity without replacing human judgement. The designer remains in charge of decisions.

AI as accelerator, not replacement

The right approach is to integrate AI into the co-design process, not use it as a replacement. Here is an effective workflow:

Phase 1: user research (human)

Conduct interviews, field observations and co-design workshops with your target users. Document insights, frustrations, spoken and unspoken needs. This phase is irreplaceable.

Phase 2: concept generation (AI + human)

From the collected insights, use AI to quickly generate dozens of concepts and mock-ups. AI accelerates this exploratory phase by proposing directions the team would not have considered.

Phase 3: testing and iteration (human)

Present the concepts to users in test sessions. Observe their reactions, collect their feedback. AI cannot replace this direct confrontation with reality.

Phase 4: refinement (AI + human)

Use AI to iterate quickly on validated concepts. Generate variations, test micro-interactions, optimise journeys. The human designer supervises and arbitrates.

Human Oversight framework

Integrating AI into co-design requires a clear governance framework. Without supervision, algorithmic biases propagate into the final product. Three principles structure a responsible approach:

1. Systematic validation: every AI-generated deliverable (mock-up, journey, content) goes through a human review before being presented to users. AI proposes, human decides.
2. Decision traceability: document which parts of the design come from AI and which result from user research. This traceability eases UX debugging and regulatory compliance.
3. User veto right: in test sessions, user feedback always trumps AI recommendations. If an algorithm suggests an optimised journey but testers find it confusing, human perception wins.

This framework is particularly relevant in the Swiss context, where the nFADP (new Swiss Federal Act on Data Protection) imposes increased transparency on the use of automated systems in processes affecting users.

Implications for Swiss businesses

A growing UX market

The Swiss UX/UI design market is growing strongly. According to Swiss Digital Initiative, more than 70% of Swiss companies plan to increase their investments in user experience by 2027. This trend is driven by the accelerated digitalisation of financial services, healthcare and public administration.

Competitive advantage through co-design

Swiss companies that maintain a co-design approach stand out on three fronts:

• Reduced abandonment rate: interfaces co-designed with users show conversion rates 20 to 35% higher than interfaces generated solely by AI, according to Nielsen Norman Group data (2025).
• Customer loyalty: in a market where customer acquisition cost is high (banking, insurance, B2B), a product that faithfully reflects real needs generates stronger retention.
• Easier compliance: documentation from co-design workshops (validated personas, tested journeys, recorded consents) provides a solid base to meet Swiss regulatory requirements.

SMEs and start-ups: an accessible lever

AI-augmented co-design is no longer reserved for large groups. A French-speaking Swiss SME can run a 5-day co-design sprint with a controlled budget: 2 days of field research, 1 day of AI generation, 2 days of testing and iteration. The result: a prototype validated by real users, ready for development.

Why this matters for Swiss businesses

In French-speaking Switzerland, the customer relationship is built on trust and proximity. A product or service designed solely by AI will always miss this relational dimension that makes the difference on the Swiss market.

The companies that will succeed in 2026 are those that combine AI's speed with the depth of human co-design. This is not a binary choice; it is a matter of orchestration.

This approach also applies to your visibility strategy. Optimising your presence for AI search engines requires understanding how your real customers phrase their questions, an insight that can only come from direct listening.

Summary

• AI generates mock-ups and prototypes quickly, but does not understand users.
• Empathy, Swiss cultural context and unspoken needs are beyond AI.
• Human co-design remains necessary to create products that resonate with the local market.
• The optimal approach combines AI (speed, exploration) and human (understanding, validation).
• The typical process: human research, AI generation, human testing, AI+human refinement.
• Discover our support offers to integrate AI into your processes without losing the human dimension.

FAQ

Can AI replace a UX designer for an SME with a small budget?

No. AI reduces production costs (mock-ups, variations, prototypes), but it does not replace user understanding. An SME with a limited budget has every interest in using AI to accelerate the mechanical phases and concentrate its human investment on user research and testing. A five-day co-design sprint with AI costs roughly CHF 10'000.– to CHF 15'000.–, against CHF 25'000.– without AI, and the result is often better.

What are the risks of a fully AI-generated design?

The main risks are standardisation (your interface looks like your competitors'), cultural bias (AI reproduces Anglo-Saxon patterns ill-suited to the Swiss market) and a gap with real needs (AI optimises for metrics, not satisfaction). On top of this comes a regulatory risk: the nFADP requires transparency on automated processes affecting users.

How do you start an AI-augmented co-design approach?

Start with a pilot project. Identify a critical user journey (onboarding, conversion funnel, customer area) and apply the four-phase workflow described in this article. Measure results (conversion rate, user satisfaction, completion time) and compare them with the previous version. Gains are generally visible from the first cycle.

Want to integrate AI into your design processes without losing the human dimension? Contact us for a free diagnosis of your user journeys and an action plan adapted to your context.