WirewalkAI Commission an engagement

The Operating Review › AI policy and governance

AI adoption

What you allow, and where it would actually earn.

The two jobs that sit either side of the build. Priced and delivered separately, and neither commits you to the other — a firm with a client questionnaire due next month needs one of these and may never want the other.

If what you actually need is the infrastructure itself, that is the other half of this practice: private AI installations, GPU clusters at any scale, and fixed-price investigations for infrastructure that is already built and slow.

Then: where it earns, and what you allow

The two jobs that sit either side of the build. Priced and delivered separately, and neither commits you to the other.

PUTTING IT TO WORK

For when you want the productivity and have nobody who can tell you where it is real.

  • Where it would actually earn. Which work in your operation is genuinely worth automating, judged against how the business runs rather than against what is currently fashionable.
  • Buy, build or neither. What a tool costs to run at your volume, what it would cost to build, and which of the obvious applications will not repay the effort. Often the honest answer is the third one.
  • Delivered so it survives you. Documented, owned by your people, and not dependent on us or on a vendor you cannot leave.

THE POLICY — $18,000

For when somebody is going to ask what you allow, and you need an answer you can stand behind.

  • What is already in use. An inventory from what the network and the expense claims show, not a survey asking people to confess.
  • A policy people will follow. Written to permit rather than forbid: named tools, what may go into each, what must never, who signs an exception. A ban is ignored by the second week and leaves you the same exposure plus a document saying you knew.
  • The evidence behind it. Processing agreements, retention terms, approvals and an audit trail.

THE GOVERNANCE — $28,000

For when a document is no longer enough, because the question has stopped being what you allow and started being who decided.

  • Every model and AI-bearing vendor, inventoried. Classified by what it touches and what it could cost you. Most of the exposure is in software you already bought, where a vendor turned a feature on.
  • Decision rights, and an approval path people use. The long one is why shadow AI exists.
  • Monitoring, an incident procedure, and a board pack. Mapped to the NIST AI Risk Management Framework and ISO/IEC 42001, and to the EU AI Act where you have European exposure — as a structure the work is delivered under, never as a certification.

Architecture and readiness assessment — $9,500

A full day with whoever owns the decision, and two days of work around it. Before the room, we establish what you are already running — from the network and the expense claims, not a survey asking people to confess. Most firms are both more exposed and further along than they think, and both halves of that change the conversation. In the room, what you are actually trying to do and what your sector will and will not permit. Afterwards, a written readiness assessment you can put to a board. The fee is credited once against a subsequent build or implementation, up to the fee due, with no expiry. Request an invoice so the credit is applied before payment. Bounded infrastructure diagnostics can be booked directly after a brief qualification call; they do not require this paid assessment.

See the assessment Ask a question first

Your sector already has a view. A law firm has bar guidance on client confidences. A defense supplier has controlled technical data that must not reach a public model at all. A practice has patient records and a business associate agreement written before any of this existed. What makes general advice usable is knowing which rules are yours, which is what the industry pages are for.