How we use AI in our research and innovation work.
We build our own AI agents and use them for one reason: to make the work better.
Our agents. Our judgment.
-
Built in-house.
Our agents, which we call Tei, are designed around how our researchers work, and run only on AI models that don’t train on client material.
-
Better work is the point.
They let us look further and check harder, across more of what people said, more markets and more of the evidence. We don’t use AI to cut the thinking short.
-
People decide.
A person works with the AI at every step. Every member of our team works at Partner or Director level and has at least 12 years of experience. They know their craft well enough to see what a tool gets wrong, and they decide what the evidence means.
Approved tools, and a person on every output.
Client material is used only in AI tools approved under our AI security policy, part of our ISO/IEC 27001-certified information security management system, on enterprise terms and never in public AI services, and every AI output is reviewed by a person before it reaches a client.
Our certification to ISO/IEC 27001:2022, from the British Assessment Bureau, covers how we manage and protect information. The researchers who review each output are responsible for the quality of what reaches you.
If your organization has its own requirements for AI, we agree how we’ll work with your material before the work begins.
What it adds to the work.
For a consumer app company, we ran fieldwork with younger users in three European markets. Between sessions, AI support let us check emerging hunches across ages and markets, and we took them back into the next sessions as sharper questions.
Choosing between three explanations of what makes younger users pay
Our thinking on AI in research
How would AI fit your project?
Ask us how we’d use it on your question, and how we’d handle your material.
A partner or director will reply.