We put a name to the judgement.
Voltage is the difference in potential between two poles: your virtue, and what your AI actually does. The advisory work is the current that runs across the gap.
Organisations are being handed obligations they were never staffed to meet: prove your people understand AI, appoint someone to own its ethics, show a regulator you have. Most governance consultancies will sell you a framework and a slide deck. We sell you a finding with a name on it: our own judgement, tested against real casework, not a framework and a slide deck.
Why we exist
Because there is a specific gap that neither a law firm nor a management consultancy nor a training vendor is built to fill. The law firm tells you what the rule says. The consultancy gives you a maturity model. The training vendor sells you completion certificates. None of them will look a board in the eye and say whether the organisation is actually competent to carry the AI risk it has taken on, and put their name to the answer. Ethics for AI, done properly, is that sentence, said by someone qualified to say it. That is the whole business.
The useful shift, in practice, is treating ethics as a capability an organisation builds, the way it builds strategy or financial discipline, rather than a compliance exercise bolted on afterwards. The operative question stops being “is this compliant” and becomes “what are we actually optimising for, and what kind of organisation does that choice make us.” That is the conversation we have with you.
Who does the work
A small, deliberately senior group. Between us we have written AI governance policy, sat as the ethics voice in the room where a launch was argued over, advised boards on risk they could not yet name, and taught the competences that the standards we work from now set out. We are practitioners who have carried the responsibility ourselves. When you engage us, the person who scopes the work is the person who does it and signs it.
We keep the firm small on purpose. Scale is how advisory firms drift: the partner sells, a pyramid delivers, and the judgement you were promised gets diluted on the way down. We would rather turn work away than dilute the thing you actually hired.
The people
Every engagement is led and signed by a named individual.
Along the way I learned how to sit with leadership and boards and tell them plainly what they needed to know, not what was easiest to hear. I am seeing the same pattern again now in AI, at a much higher cost if we get it wrong, and that is exactly the skill ethics for AI needs: someone who can translate a technical risk into a decision a board can actually act on.
This practice exists because that gap, between what a technology can do and what an organisation can actually stand behind, does not close on its own. Someone has to insist on it. Not a compliance checklist or an abstract lecture on ethics, but a straight conversation about where your business is exposed, what it will cost you, and what to do next.
We keep the bench small and senior on purpose. When an engagement needs a specialist we don't have, we say so and bring one in named, rather than pretend the range in-house.
What we believe
- An ethicist without authority has no way to act on what they find. A judgement with no power to change a decision is just an opinion, however correct, and an organisation that only wants the opinion is not actually asking the ethical question. Get the authority wrong and nothing else on this list matters: the best judgement in the world is worthless to a board if the person holding it has no way to make it count.
- Competence is measurable. "AI literacy" is not a vibe. Named, external competence standards exist for the role, and evidence can be built against them or it cannot. We build it.
- Independence is structural, not a promise. You cannot trust a finding from someone paid by the party it favours. So we simply are not paid by them. The wall is in the invoices, not the values statement.
- The name on the page is the point. Anonymous frameworks let everyone off the hook. A signed finding is one a person will defend, and one a board can rely on.
- Legal and ethical are not the same test. An organisation can tick every compliance box on the page and still make a call it would struggle to defend to the people that call affects. We are hired to answer the second question, not just confirm the first.
- The floor is moving, on purpose. Responsible AI set the baseline: minimise harm, meet the standard, don't be the case study. That baseline is necessary and it is not the ceiling: a system can clear it entirely and still be the wrong thing to have built. The field's own direction of travel calls the next stage AI stewardship, judging a system not only by what it avoids doing wrong, but by what it is actually for, and whether that purpose is one worth pursuing for the people who will live with it. We build against wherever the floor sits today, and we treat stewardship as the direction it is moving in, not a future someone else gets to define.
What we will not do
- We will not dress up someone else's tool as independent advice. The only things we sell are our own, under our own name.
- We will not soften a finding to suit a client, or dress an open-ended retainer up as strategy.
- We will not sell a board a role design we know has no teeth.
- We will not pretend to certify anyone. We advise; we do not issue credentials.
Virtue without voltage is a mission statement with no current running through it. Voltage without virtue is a business that moves fast and asks nobody. We're the bit in the middle that insists on both, every time.