Boards are accountable for AI risk before they understand the stack. Twelve questions that turn an opaque program into a governable one without slowing it down. Thought Leadership by Stratenity Advisory Team.
Why this matters now
Boards are increasingly being asked to sign off on AI strategies they cannot fully evaluate. The risk is real on both sides. Under-governance invites failure modes that hit headlines. Over-governance freezes programs that need to ship to compound. The middle path is a small set of disciplined questions, asked every quarter, that surface the right information without forcing the board to legislate technical detail. Boards that ask these questions consistently raise the quality of management's answers within two cycles, because the discipline of preparing for the questions reshapes the program itself.
Our point of view
Twelve questions in four groups give a board the visibility it needs. The groups cover portfolio, risk, people, and compounding. Each group is small enough to discuss at a single board meeting and substantive enough to surface what is actually happening in the program.
Portfolio
The first three questions establish what the program is doing and where the evidence lives. What three use cases create the most enterprise value this quarter, and what is the evidence behind that ranking. What pilots have been retired this quarter, and what triggered the decision. What is the unit cost trajectory of the most-used AI workloads, measured honestly across inference, retrieval, evaluation, and override. A program that cannot answer these three questions has not yet reached operating maturity.
Risk
The next three questions establish how the program is bounded. Which AI workloads touch regulated data, and what controls govern them. Where would a model failure cause material customer or financial harm, and what is the mitigation. What is the incident-response runbook when an AI system misbehaves, and when was it last exercised. The honest answers to these questions reveal which risks the program is actually managing and which it is hoping not to encounter.
People
The next three questions establish accountability. Who owns the AI platform end to end, with named budget authority. Where has AI changed how decisions get made in the operation, and who is accountable for the quality of those decisions. What skills gap is being addressed this year and how. Programs that cannot name a platform owner are programs that will be re-organized within twelve months.
Compounding
The final three questions establish whether the program is compounding or merely spending. What did the team reuse this quarter that was built last quarter. Which capabilities will be standard across the firm in twelve months. What would a 25 percent increase in AI investment buy, and what would a 25 percent decrease cost. The honest answers to the compounding questions reveal whether the program is an asset or a recurring cost.
What to change this quarter
Publish the twelve questions as the board's standing AI agenda. Require management to come with evidence, not slides. The conversation level rises immediately, and so does the quality of the decisions that follow.