Warm light glowing through layered translucent paper screens — a sense of quiet enclosure.

Where your data goes

The question every partner is really asking — answered before you ask it.

For years the objection to AI was whether it worked. That argument is over. The real objection now — the one a principal actually loses sleep over — is simpler and sharper:

“Where does our data go?”

Most vendors treat that question as an obstacle to get past. We treat it as the first thing to settle. Every tool we prescribe arrives with its data posture written down — not implied, not “enterprise-grade,” not buried in a policy nobody reads.

This is not caution for its own sake. It is the habit of fifteen years handling confidential matters at the highest professional level, brought to a place it was badly needed.

Three questions, every time

What “in writing” actually means.

For every tool that could touch your work, the prescription answers the same three questions. You never have to take the posture on faith.

Where does it travel?

Which servers see your information, in which jurisdictions, and who else can reach it. If a tool sends your files somewhere you would not want them, you learn that before it touches a single document.

Does it train on your data?

Whether your information is retained, used to improve someone else’s model, or shared onward — stated plainly, per tool, in writing. Defaults change; we track them so you do not have to.

What did we rule out — and why?

The tools we rejected on data grounds alone, named, with the reason. The absence of a recommendation is itself part of the answer, and you get to see our reasoning.

And a line we will not cross

AI drafts. Your people send.

Data posture is one half of trust; restraint is the other. AI never speaks to your clients unsupervised — not now, and not as some future feature. It prepares the work and a human makes the call. Where a system runs on its own, a person still stands at the checkpoint. We hold this as doctrine, and we would not want a client who wanted otherwise.

  • AI drafts; a person reviews and sends.
  • No client-facing message goes out unsupervised.
  • Every automated system keeps a human checkpoint.
  • The posture is written down, not promised.

Bring us the workflow you would never let a black box near.

On the free call we will tell you, out loud, where the data would go — and whether we would touch that workflow at all.

Book the diagnostic call

12 of fifteen advisory seats open