Our approach

From first conversation to lasting safeguards.

Every engagement is shaped around your organisation, but the journey follows the same four steps. Each one leaves you with something useful, even if you choose to stop there.

  1. 01 · Listen

    A first conversation

    We start by understanding what you are doing with AI, what you are planning, and what concerns you. It is a conversation, not a sales call.

    By the end you will know whether we can help, what that would involve, and what you could do on your own.

    What you leave withA shared, written understanding of the question to be answered.

  2. 02 · Map

    Understand the landscape

    We find out where AI is in use, what it can reach, and who relies on it, including the tools that arrived without a formal decision.

    Then we assess what could go wrong, how likely it is, and how much it would matter if it did.

    What you leave withA clear picture of your AI risk, ranked by what matters most.

  3. 03 · Safeguard

    Design and test the defences

    We work alongside your teams to put the right controls in place: permissions, limits, human checkpoints, monitoring and training.

    Then we test them adversarially, the way an attacker or a misbehaving model would, to make sure they hold up under pressure.

    What you leave withSafeguards that have been tried, not just written down.

  4. 04 · Sustain

    Keep them working

    AI systems change, and so do the ways people try to misuse them. We help you build safety into how you already work, with clear owners and simple, regular reviews.

    The aim is for your own people to spot new risks early and know what to do about them.

    What you leave withLasting capability in your team, not dependence on ours.

Working with us

What you can expect.

  • You work with the people doing the work

    The person you first speak to stays involved throughout. No hand-offs to someone you have never met.

  • Findings you can act on

    Every recommendation is written in plain English, with a clear reason and a practical next step.

  • Proportion, not alarm

    We tell you when something is fine. Not every risk needs a control, and we will not invent work.

  • Discretion as standard

    What you share with us stays with us. We are happy to work under a confidentiality agreement from the first conversation.

Step one

It starts with a conversation.

No preparation needed. Tell us where you are with AI, and we will help you work out what to do next.

Start a conversation