How we run our own practice.

Our core conviction is simple to state and hard to live: AI can accelerate the work, but it cannot carry the responsibility for the outcome. A machine can draft, classify, predict, and surface patterns. It cannot hold judgment, intention, or accountability. When organizations forget that distinction, responsibility scatters across tools, vendors, and workflows until no one can say who was accountable. Our work exists to keep that from happening. So does the way we work.

These are the commitments we hold ourselves to.

We keep accountability with a person.

Every recommendation we make, and every use of AI in our own work, has a named human answerable for it. We do not let a tool become the reason no one is responsible.

We match the level of judgment to the level of consequence.

Low-stakes work can move quickly. High-stakes decisions require human review proportional to what is at risk. We help clients set that threshold deliberately, and we hold to it ourselves.

We separate what we know from what we think.

In our advice, our writing, and our public commentary, we distinguish evidence from interpretation and recommendation. We do not dress opinion as fact, invent statistics, or overstate what the research supports. If we are uncertain, we say so.

We tell clients what should not be automated.

Our value is not in helping you deploy AI everywhere. It is in helping you see what should be accelerated, what should be protected, what should stay human, and who remains responsible. Sometimes the right counsel is to slow down or not to automate at all. We will give it.

We treat the people affected by a system as participants, not obstacles.

The workforce living with an AI decision, and the community on the receiving end of it, have a stake in how it is designed. Good governance brings them in. We advise that, and we practice it.

We are transparent about limits — including our own.

We name what a tool cannot do, what our own analysis does not cover, and where a client should seek other expertise. Trust is built on disclosed limitations, not hidden ones.

We protect what clients entrust to us.

What we learn inside an engagement stays there. We do not trade on confidential information, and we do not use one client's situation as material for another without permission.

We refuse work that requires us to compromise these commitments.

Some engagements ask a governance firm to provide cover rather than scrutiny. We decline them. Governance has to be operational, not decorative — for our clients and for us.

The discipline behind the statement

The future of work is neither fully human nor fully machine. It is deliberately designed. Ethics is not a statement we post and move past; it is the discipline of deciding, again and again, what to accelerate and what to protect. We hold ourselves to that standard so we are worth trusting with yours.

Questions about how we work, or how these commitments apply to an engagement, are welcome at contact@humanmachineworks.com.