How we use AI

AI is reshaping how many organisations work, and we believe it has genuine value. But that value must be earned carefully. As a research agency whose whole purpose is helping people understand real human behaviour, we have a particular responsibility to be clear about where AI fits in our practice, and where it does not.

Where we use AI

We use AI to make our business and our process more efficient in places where it genuinely supports the team without compromising the quality or integrity of our work.

• Business development and administration, including drafting, scheduling, and internal communications
• Research planning, such as initial desk research and early-stage preparation
• Transcription and data organisation to help researchers work more efficiently with raw material
• Research analysis, as an assistant only. AI may help surface patterns or organise data, but all interpretation is led by experienced researchers

Where we will always be human

The core of our work will always be delivered by humans with humans.

Scoping the project

Understanding a client’s problem deeply, defining the right research question, and choosing the methodology that will genuinely answer it requires contextual, relational, and critical thinking that AI cannot replicate. Our approach is bespoke because it has to be. Every project begins with us getting properly under the skin of what a client needs, and that demands human judgement and human conversation.

Running research 

Our core research services will always be led by human researchers and conducted with real people. Human behaviour and attitude are nuanced, contradictory, context-dependent, and shaped by lived experience that cannot be fully captured in historical data that feeds LLMs. That is precisely why user research exists. It is the mechanism through which we discover what no amount of AI processing can tell us.

Human-first analysis

It is very easy to be confidently wrong with AI. Language models are trained on historical data. Much of what user research uncovers is genuinely novel: attitudes, experiences, and behaviours that have never previously been captured, and which exist precisely because we went looking for them.

Identifying what research themes really mean, how they connect, and what they should tell a client about their service is something experienced researchers must lead. AI can assist with the mechanics of analysis. It cannot own the meaning-making.

This principle also matters for the development of our team. The critical thinking skills that make a good researcher take time and practice to build. We will not shortcut that by handing interpretation to AI and removing the conditions in which those skills can grow.

AI in the research ecosystem

There are new tools that use AI to run research and AI that can act as research participants. We are watching this space carefully and with open minds.

Where these tools can play a useful supporting role, whether pressure-testing a proposed approach, exploring initial hypotheses, or helping scale certain elements of a project, we will consider them as part of our process. But they will always sit alongside our core work, not replace it. The distinction matters to us, and we believe it should matter to our clients too.

This thinking also shapes CollectivAlly, the platform we are building alongside Web Usability. CollectivAlly combines AI with ethically sourced, paid lived-experience insight to help teams explore accessibility and usability barriers earlier in the design process. It is not about replacing real users. It is about amplifying their voices and making inclusive thinking more accessible to organisations of all sizes. The same principle applies: AI as a tool that expands human understanding, never one that replaces it.

Our lines in the sand

To be clear about our commitments:

We will

• Use AI to make our team more efficient and our process sharper, where it genuinely helps
• Use AI as an assistant in analysis, to support researchers in organising and exploring data
• Engage thoughtfully with new AI tools in the research space and use them where they add genuine value
• Be transparent with clients about where and how AI is used in their projects

 We will not

• Use AI to define what a client needs or determine how to answer their research question
• Allow AI to lead on interpretation, or treat AI-generated analysis as a substitute for experienced human judgement
• Totally replace real research participants, or real research conversations, with AI-generated alternatives
• Use AI to cut corners on the quality of insight we deliver, or the dignity of the people who contribute to our research

Questions about how we use AI on a specific project? We are always happy to talk through our approach so please get in touch.