01The Board's AI Mandate
What changed, what are we accountable for, and what does good look like?
Board keeps: Board AI Position Statement
10 modules, 24 hours.Every one ends in something the board keeps.
Each module closes with a document the board adopts: a position statement, a charter amendment, a reporting standard, an investment protocol. Nothing is delivered that the board cannot use at the next meeting.
90 minuteslive board session
60 to 90 minutespre-work: reading and a podcast episode
1 breakoutfacilitated, groups of 4 to 6
1 artifacta decision document the board keeps
01
02The Language of AI
Can every director follow the conversation without translation?
Board keeps: Director glossary and 25 questions for management
03Where We Actually Are
How mature is this company, honestly?
Board keeps: AI maturity diagnostic and gap map
04Agentic AI and the ROI Question
Where does money get made, and how do we tell real ROI from theater?
Board keeps: Agentic value map and ROI challenge protocol
05The Economics of Intelligence
What does this cost at scale, and who is watching the meter?
Board keeps: AI cost reporting standard
06Build, Buy, Tune or Rent
Should we own models, data or infrastructure?
Board keeps: Model strategy decision memo
07Defending the Enterprise
What does an AI-enabled attack look like, and are we ready?
Board keeps: Threat tabletop and incident checklist
08Governance and Fiduciary Duty
What is our legal exposure, and what record are we creating?
Board keeps: Charter language and reporting cadence
09The Competitive Board
What are competitors doing that we cannot see from here?
Board keeps: Competitive AI intelligence brief
10Becoming an AI Native Board
How do we change the way this board works?
Board keeps: 12-month plan and certification submission
Specialist tracks for the sector you govern.
3 hours each, added to the core program.
Physical AI for industrials
Robotics, vision systems, humanoid pilots, OT security, capex sequencing, workforce and safety governance.
Regulated industries
Model risk management, clinical and claims decisioning, fair lending, EU AI Act Annex III exposure.
Private equity boards
AI in the thesis, the 100-day plan, portfolio platform decisions, diligence and the exit narrative.
The board's own AI
Using AI on board materials without creating a privilege, retention or discoverability problem.
The CEO and the Board on AI
A 90-minute executive session on aligning ambition, pace and risk appetite.
AI and CEO Succession
A 90-minute executive session on what the next CEO must be able to do.
The Nominating Committee's AI Problem
A 90-minute executive session on composition, advisory boards and the director market.
Inside Module 1: The Board's AI Mandate.
What changed, what the board is accountable for, and what good looks like.
Capability crossed a threshold
Systems now do work that required a person, without being reprogrammed for each task. Software that executes rules became software that handles situations.
The interface became general
The same system serves customer service, contract review, code, forecasting and claims. It does not live in 1 function, so it cannot be governed in 1 function.
Systems started acting
Agents plan multi-step work and call tools. An answer is advice. An action is an act of the corporation.
What has not changed: the technology is probabilistic. It gives its best guess confidently, including when the guess is wrong. Every governance question traces back to that.
| Management owns | The board owns |
|---|---|
| Which models, vendors and architectures | Whether the strategy is coherent and ambitious enough |
| Use case selection and sequencing | Risk appetite, and whether it is stated |
| Technical controls and evaluation | Whether an oversight system exists and works |
| Delivery, integration, change management | Capital allocation and stage gates |
| Day-to-day AI incident response | Escalation thresholds and readiness |
| Building the AI talent bench | CEO capability, succession and board composition |
The test: the board's job is not to know how the system works. It is to know whether the company knows, whether the board would be told if something went wrong, and whether it is asking for the right things.
1996
In re Caremark
Establishes directors' oversight liability.
2006
Stone v. Ritter
Grounds oversight in the duty of loyalty.
2019
Marchand v. Barnhill
Mission-critical risks need more rigorous oversight.
2021
Boeing
The derivative decision sharpens the standard.
2023
In re McDonald's
Extends oversight duties to officers.
Why it matters for AI: AI becomes mission-critical wherever it decides credit, hiring, pricing, claims or disclosure. AI systems also produce clean-looking reports while behaving badly, so the board's record matters more. Ask what the last 4 meetings' minutes say about AI.
Pilot theater
Impressive demos that never reach production.
Tell: reports count pilots launched, not workflows changed.
Misallocation
Budget goes to visible front-office uses while returns sit in operations.
Tell: the portfolio follows glamour, not cost and friction.
Broken process
AI layered on a bad process gives a faster bad process.
Tell: nobody describes the workflow after AI, only the tool.
Going it alone
Vendor-partnered builds succeed about twice as often as internal-only ones.
Tell: a build justified by control, with no time-to-value comparison.
90% vs 40%
Employees at over 90% of organizations use personal AI tools at work, while about 40% of firms buy enterprise subscriptions. The first question is not whether to adopt AI. It is what is already happening.
Board AI Position Statement
1 page, adopted by the board, reviewed every year. The document the chair hands to the CEO.
- What AI means for this business, in the board's own words
- Our ambition: fast follower, sector leader, or selectively aggressive
- Our risk appetite: where AI may act without human review
- What the board requires, and on what cadence
- Who oversees it, and how committees interact
- What would change our mind
The breakout: where boards find the gap themselves.
Management is absent from breakouts and present for the report-back.
What do we believe?
Position8 min
Pressure5 min
Proof4 min
Ask3 min
1Groups of 4 to 6. Never the whole board in 1 conversation.
2Deliberately mixed. Never all the technologists together.
3Every group ends with a written position, read aloud in under 90 seconds.
4The facilitator surfaces disagreement instead of resolving it. Boards that agree too quickly on AI have not understood the question.
5Management returns for the report-back. This is the most valuable structural choice in the program.