AI DECISION GUIDE
What should a board ask before funding AI?
Before funding AI, a board needs a defensible mandate: a business objective, an accountable executive, explicit authority limits and evidence for the next investment decision.
Eight decisions for the board agenda
- Ambition: what outcome justifies AI rather than an improvement to the existing process?
- Investment: what funds discovery, a pilot and production? What remains conditional?
- Autonomy: does the system recommend actions or execute them? On whose authority?
- Sourcing: what should be bought, built or delivered with partners?
- Governance: who approves, monitors and can suspend the system?
- Workforce: whose work changes, and who owns adoption?
- Foundations: are data, access rights and integrations dependable enough?
- Risk: what triggers escalation, a stop or independent review?
The minimum investment pack
Request a concise record for each use case: a named sponsor, current process and baseline, target KPI, total cost, delivery timetable, data scope, autonomy level, risk owner and decision gate. A compelling demonstration does not establish these conditions.
Separate immediately approved funding from conditional funding. A board can authorize a bounded experiment while making production contingent on measurable value and adequate controls.
When the evidence is missing
A deferred decision is useful when it identifies the evidence required, its owner and a due date. Broad approval can otherwise transfer undefined risk to delivery teams.
Use the Board Kit to align these positions. Once the mandate is clear, use the Navigator to test one specific investment case. Both outputs belong in the same funding discussion.
Start here: the first diagnostic question
Your answer carries into the tool in this browser. No email is needed to start.
Connect governance and value
The Board Kit clarifies the mandate. The Navigator examines one use case’s value and readiness.
Sources and limitations
These frameworks inform governance and risk analysis. They do not validate ApexTransform scores or financial coefficients. Methods and examples are ApexTransform analysis, without endorsement by these organizations.
FAQ
Should a board review every model?
Set reserved decisions, delegation rules and escalation thresholds. The required depth of review depends on risk and the mandate.
Can exploration start before the ROI is proven?
Yes, with a limited discovery budget, a testable hypothesis and an explicit gate. Production funding requires suitable evidence.
Analysis: Stéphane Gervais, ApexTransform · Updated 19 October 2026