Role-Immersion · Public Company · Branching D&O simulation
The board pack asks for full production approval of an agentic AI system. The CEO has already told investors it exists. Legal hasn't reviewed the claims. You have 48 hours, and none of the three votes is safe.
The situation
You receive a board pack from the Company Secretary: Agentic AI System, Full Production Approval. It covers pilot results, projected ROI, a vendor assessment and a recommendation to proceed. The financials are attractive. The risk section runs two paragraphs. Nobody has independently validated the model, the vendor assessment cites “proprietary training data” without addressing data rights, bias testing or IP ownership, and the 85% confidence threshold comes with no methodology.
The company is already talking about the system in public. The website says it is “transforming customer outcomes through intelligent automation,” the CEO referenced it in last quarter's investor briefing, and the latest analyst pack carries AI talking points. Legal has not reviewed any of it.
In the boardroom, four executives give conflicting signals. The CTO reports strong pilot metrics and not a single escalation. The CFO says new business is tracking 30% below forecast and deals are being lost to competitors' AI. The CEO does not want to explain a pullback to analysts. The General Counsel notes the D&O insurer's renewal questionnaire now asks about AI governance, and the board has not reviewed the draft answers. The Chair reminds the room that the decision will be minuted and that each director is personally accountable.
“You were on the Risk Committee. You reviewed the board pack. You voted on this. Looking back, was our process defensible?”
The Chair · Emergency board meeting
The decision on the table
The Green Light runs live with your directors, with the board pack, the boardroom scene, both branches and a debrief built from your board’s own choices.