Co-design is based on a simple but powerful idea: services and campaigns work better when the people they are intended for are involved in making them.

In an age when organisations are increasingly looking to AI for faster insight into human attitudes and behaviour, this principle matters more, not less. AI can help identify patterns, summarise data and accelerate analysis, but it cannot replace the value of talking to real people about their views, motivations, barriers and everyday lives.

Over the past two decades, co-design has moved from the margins into mainstream public policy. However, the speed and cost effectiveness of AI, may threaten its continued growth. This would be a huge blow as I’ve seen at first hand how powerful it can be when delivered well: not as a token consultation exercise, but as a disciplined way of understanding people’s needs, testing assumptions and designing around real lives.

One of my agency career highlights was working with young people in Kent to create HOUSE, an award-winning pop-up youth hub designed to help young people make positive choices about their health and wellbeing.  The space felt more like a mate’s house than a formal service. It offered DJ sets, dance machines and relaxed activities alongside support on issues such as smoking, sexual health and emotional wellbeing. Because young people felt safe and listened to, they were more willing to talk about their concerns with trusted staff.

One thing’s for sure – Chat GPT would not have generated this innovative cross-sector solution.

What does the evidence say about the value of co-design?

Individual case-studies matter, but they only take us so far. To look more closely at the evidence — and at why human-led insight still matters when so much insight work is now automated — my agency recently carried out a rapid review of UK research into co-design.The review asked two main questions: does co-design lead to better public services, and what does high-quality co-design actually look like?

Behind both questions sits a bigger challenge for organisations: if we want to understand human behaviour, we cannot rely solely on data models, dashboards or AI-generated assumptions. We still need to involve the people whose lives, choices and experiences we are trying to understand.

 

The headline finding: co-design works, and gives organisations insight AI cannot provide on its own

The review found encouraging evidence. People who take part in co-design often gain confidence, skills and a stronger sense of control. It can also build trust between communities and organisations, which in turn can encourage people to use services that could make a real difference to their lives.

Just as importantly, it helps organisations hear the nuance behind behaviour: what people are worried about, why they may not trust a service or product, what they would actually use, and what would make change possible.

Across different sectors, the review found promising examples:

• Smoking cessation campaigns developed with local people can increase people’s intention to quit.
• Mental health programmes shaped by people with mental illness gave purpose, improved confidence and increased use of support services.
• Culturally tailored COVID-19 campaigns for Black communities helped build trust and improved uptake of vital public health advice.
• In community safety, co-produced approaches have been linked with reduced fear of crime and better understanding of local concerns.
• In schools, involving pupils and parents can improve attendance, motivation and relationships with the wider community.

However, the review also found an important gap. There is still limited causal evidence linking co-design directly to harder outcomes such as lower crime, fewer hospital admissions or long-term cost savings. That does not mean co-design is ineffective. It means public services are complex, and co-design is usually just one factor in delivering positive change. To understand its full impact, we need better-funded, longer-term evaluation.

We also need to recognise that insight into human behaviour is not just about what can be inferred from data; it is also about what people say, feel, notice, avoid, value and do in context.

What does good co-design look like?

The research is clear that meaningful co-design is very different from a quick consultation exercise or an AI-generated persona. The strongest projects tend to share five features:

• Shared power between professionals and the public.
• Early involvement of communities in defining the problem, not just reviewing expert-led solutions.
• Ongoing participation in delivery and evaluation, not just design workshops.
• Testing and refining ideas with people before rolling them out widely.
• Long-term funding and leadership support so the work can move beyond one-off engagement.

We have recently used this evidence to refine M‑powered, our bespoke co‑design service combining psychologist‑led focus group work, opinion polling, ethnography, as well as online and real-life community and audience panels.

The future: AI should support community intelligence, not replace it

Co-design is already improving the experience, accessibility and quality of services across areas such as health, social care, education and community safety. It also shows real promise in helping people make positive changes, from stopping smoking to getting the right mental health support earlier.

This is why organisations should be cautious about treating AI as a shortcut to understanding people. AI can be useful, but it should strengthen community intelligence, not replace it.That is why community intelligence — CI — matters. If we want services, campaigns and interventions that people understand, trust and use, organisations need to keep talking to real people and involve them in decisions from the start. AI may help us process what we learn, but co-design helps us ask better questions, challenge our assumptions and stay connected to the behaviours, beliefs and realities that shape people’s choices.

The conclusion from the review is encouraging but realistic: co-design can make services better, fairer and more trusted. The next challenge is to invest in it properly, evaluate it robustly and make sure community voices remain at the heart of public service change. In the age of AI, that is not an old-fashioned view. It is the safeguard that keeps insight human, grounded and useful.

Ed Gyde