Exploring the role of AI in our design process at Co-op
At Co-op, our designers are embedded across different product teams, solving very different problems. That makes it valuable, and necessary, to step back and learn together.
We recently brought the whole design community together for a Design Day focused on how we should use AI in the user-centred design process, and where we should be cautious.
The topic for Design Day 2026: AI in the user-centred design process
We used Design Day as a way to move quickly from curiosity and learning to practical exploration.
We designed the day to test how AI could (and could not) support real design problems.
The structure of Design Day
Design Day was structured in a similar way to our previous Service Jams. This format helps us rapidly immerse ourselves in a problem space and then prototype solutions to problems within a few hours.
We put together teams, mixing up skills and experience from people who don’t normally work together.
We started with a kick-off to agree on how we might use AI in our design process.
We wanted to understand how AI could:
- improve the speed and quality of our design workflows
- help us better understand user needs and behaviours
- support rapid prototyping and idea exploration
- reduce repetitive tasks so we can focus on craft and problem solving
We also wanted to be honest about:
- where AI introduces risk or bias
- what “responsible use of AI” looks like for Co-op
- how confident (or uncomfortable) we feel using it
Lightning talks helped to focus our thinking
We’d arranged for some of the designers to prepare short presentations on how they’re currently using AI in their work. We asked the teams to write notes on the problems and opportunities they spotted. The notes could then form the basis for ideas the teams would work on in the afternoon.
Our talks were on:
- Using AI for skill gaps
- AI and content design: Please use responsibly
- Building and deploying an app in a day
- Designing when you’re not a designer
- Exploring AI in the design process
Team ideation
After a break the teams had time to talk through their notes, pulling out interesting ideas, sketching and voting to help decide what problem they might want to use AI to work with in the afternoon.
Rapid prototyping
During the afternoon teams could use AI to try to solve the problem they’d identified. This could be using AI to:
- generate a plan
- bring information together on your idea
- create a user interface
- prototype a workflow
- create a conversational prototype
Teams could also try using traditional design techniques to try and solve an AI problem or evolve an AI opportunity.
Team presentations
In the last hour each team presented back. We asked them to focus on:
- the idea they had
- the prototype or process
- what AI(s) you used
- what you learned
- what you could do next
Experience atlas
An AI brain that helps you to avoid duplication by finding out which solutions have been already been tested or developed.
Misson patch collection
A digital way to collect mission patches (stickers), that you can always access, long after you’ve changed laptop.
Coop-E: Digital sustainability helper
An AI plug-in tool to help you manage your digital footprint and reduce digital waste. It’s name is Co-op-E, inspired by Wall-E.
Stakeholder persona bot
An app that helps you to get feedback before presenting ideas to stakeholders. You can also pre-empt questions to help you prepare and make adjustments.
End-of-year review wrapped
A way to collate everything you’ve done in the last 12 months for end-of-year reviews. Everything from your calendar. Miro, Sharepointand Slack in one place.
AI decision guide
Helps you choose when and how to use AI. It balances the benefits of using AI with energy use and space for critical thinking.
What we learnt
Bringing the whole design team together gave us a shared, practical understanding of how AI is already useful and areas we need to explore further.
AI is most useful when it accelerates, not replaces, design thinking
Across teams, AI worked best when it supported clear problems and intent. It helped us:
- get from 0 to 1 quickly with ‘good enough’ outputs
- explore more ideas in less time
- make sense of complex or messy information
This was particularly valuable in early-stage thinking, prototyping and iteration.
However, AI struggled to define the right problem or direction. A pattern that emerged was that AI helps with some of the design creation, but we need the design process to make sure we’re solving the right problem and making human-centred decisions.
It improves speed, but not always confidence
AI saved time on repetitive and routine work, and helped teams move faster across:
- idea generation
- prototyping
- synthesising information
It also lowered the barrier to trying things, making experimentation feel more accessible across the team, regardless of role or experience.
However, trust in outputs is still a challenge. Teams often found that:
- outputs could be inconsistent or misaligned
- content lacked our voice
This means speed gains are only valuable if we also build confidence in how we use AI and the results it provides with.
Prompts, context and craft matter
Another takeaway was how important prompting is. High-quality outputs depended on:
- clear intent
- structure
- iterative refinement
Without this, outputs became vague, overwhelming or misleading. This reinforced that using AI effectively is a skill, not a shortcut to design expertise, planning or critical thinking.
AI is already changing how we collaborate
Several teams found that AI:
- helped bridge gaps between disciplines (for example, design and engineering)
- enabled more interactive and dynamic prototypes
- made it easier to communicate ideas visually and quickly
It also created new opportunities for collaboration, particularly when:
- sharing prompts
- iterating on outputs together
- using AI as a shared “thinking tool”
At the same time, it could silo teams, with typically only one person prompting and collaborating with AI. How we could use AI as a member of a larger group is something we want to explore further.
Our current ways of working aren’t fully set up for AI
AI doesn’t yet fit neatly into our existing processes. Teams experienced friction around:
- where AI fits within the design lifecycle
- how it works with our Experience Library design system
- limited access to tools or environments
There’s also a risk of:
- over-reliance on AI
- AI creating surface level outputs that appear finished but lack design expertise, standards and critical thinking
This highlighted the need to deliberately evolve how we work, rather than slotting AI to into existing ways of doing things or replacing key, human parts of the design process.
We must use AI responsibly
Alongside the opportunities, teams raised important concerns around:
- accuracy and bias in outputs
- overconfidence from AI in incorrect outputs
- environmental impact and energy use
- whether AI is always the right tool for the job
There was recognition that just because we can use AI, doesn’t always mean we should.
Design Day helped create space to talk about this openly, and reinforced the importance of using AI responsibly, for both our customers and our colleagues.
What’s next
Design Day wasn’t about reaching a final answer on AI. It was about creating a shared understanding around where we are. We’re now focusing on turning what we learnt into practical, scalable ways of working.
Turning experimentation into shared practice
We’ll continue to experiment within product teams and a new AI focused community of interest with an emphasis on:
- sharing patterns, tools and examples that others can reuse (prompt templates or shared libraries)
- capturing what works well (and what doesn’t)
- guidance on when to use AI and when to use traditional methods
- standards for responsible use, including validation, environmental impact and fact checking
- better integration of AI into our design workflows
- adapting our design systems to work with AI-generated outputs
- creating safe “sandbox” environments for experimentation
This will help us move from individual experimentation to team-wide capability.
Design and AI are evolving quickly. We need to continue to evolve our view on AI tools and their value in user-centred design.
Blog post by Matt Tyas, Head of Design
Find out more about topics in this blog: