Exec AI Literacy Program

Use this when a board, C-suite, or senior management team needs to get fluent in AI fast, and someone has confused that with teaching them to use ChatGPT. The job is to build a program that lifts the quality of the decisions these leaders actually make: where to invest, what to govern, what to believe, where AI fits the business. This is not a tools course. If the request is really an org-wide capability rollout for practitioners, use /training-curriculum-architect; if it is an assessment of where the org stands, use /ai-maturity-org.

Related skills: Pairs with /training-curriculum-architect for the practitioner layer beneath the exec layer. Grounds the program in a maturity read from /ai-maturity-org. Sits inside the broader mandate in /fractional-caio-playbook, and structures individual leader development with /learning-path.

The hard part most teams miss

Executives do not need to learn to prompt. They need to make good decisions about AI. A board member will never write a production prompt, but they will approve a budget, sign off on a governance policy, and decide whether to believe a vendor's accuracy claim. Teach the judgment those decisions require: how to size an AI bet, how to read a risk, how to tell a real capability from a demo. A literacy program built around tooling trains the wrong muscle for the people in the room.
Literacy without hands-on does not stick, and the failure mode is over-trust. A leader who has only seen AI work in a polished demo will trust it in places it has no business being trusted. The fix is not more slides. It is putting leaders in front of a model and letting them watch it confidently get something wrong, on their own data, on a decision they recognize. Calibrated trust comes from seeing the failure, not from being warned about it.
The goal is better decisions, not certificates. Attendance, completion rates, and satisfaction scores measure whether people showed up, not whether the program changed anything. The only outcome that matters is whether the next AI decision this team makes is better than the last one. So instrument the decisions: tie the learning to a real choice each leader faces, and measure the choice.

Process

Step 1: Gather inputs

Ask the user:

Who is in the room? ({{audience}}: board, C-suite, senior managers, or a mix. Each tier makes different decisions and needs a different depth.)
What AI decisions are these leaders facing in the next two quarters? ({{live_decisions}}: a specific investment, a governance policy, a build-vs-buy call, a vendor evaluation, a workforce question. The program should attach to real ones.)
What is the organization's current AI maturity? ({{maturity_level}}: skeptical, experimenting, scaling, or already burned by a bad bet. This sets the starting altitude. If unknown, run /ai-maturity-org first.)
What is driving the ask? (A board mandate, a competitor move, a failed pilot, a regulatory deadline, fear of being left behind. The driver shapes what "good" looks like.)
What format and cadence is realistic? ({{format_constraints}}: a single offsite, a recurring series, embedded in existing leadership meetings. Senior calendars are the binding constraint.)
How will success be judged, and by whom? (Name the person who decides this worked, and what they will look at. If the honest answer is "attendance," that is the first thing to fix.)

Step 2: Map each tier to the decisions it owns

Different seats make different calls. Do not teach the same thing to all of them.

Board: oversight and risk. They must understand AI capabilities and real limits well enough to challenge management, judge whether the company's AI strategy is sound, and govern risk and accountability. They do not need implementation depth. They need to ask the right hard question.
C-suite: strategy and investment. They must judge ROI and opportunity cost, decide where AI fits the business model, set the governance posture, and own the trade-offs between speed and safety. They need enough fluency to not be sold by a vendor or stalled by a skeptic.
Senior managers: translation and execution. They must connect AI capability to their function's real work, spot where it helps and where it is theater, and lead teams through adoption without over-promising. They need the most hands-on of the three tiers.

For each tier present, write the two or three decisions they actually own. Those decisions are the spine of their curriculum.

Step 3: Set objectives as decisions, not topics

For every tier, turn each owned decision into a learning objective phrased as a capability. Not "understand machine learning," but "evaluate a vendor's accuracy claim well enough to know what to ask for." Not "learn about AI risk," but "decide which AI uses in our business require human sign-off." Topics are forgettable. Decisions are testable.

Mark anything you are inferring rather than confirming with .

Step 4: Design the modules

Build modules that serve the objectives, drawn from this core literacy set and pruned to what each tier needs:

Capabilities, honestly. What current AI does well, with examples from the org's own domain. No hype, no doom.
Real limits and failure modes. Where it breaks, why it breaks, and what a confident wrong answer looks like. This is where the hands-on work lives.
Risk and governance. Data, privacy, bias, accountability, the regulatory surface relevant to this business. Who owns what when AI is in the loop.
ROI and investment judgment. How to size a bet, read a vendor claim, spot opportunity cost, and tell a pilot worth funding from a science project.
Where AI fits this business. The strategic map: which parts of the operating model AI changes, which it does not, and where the company is exposed if it moves too slow or too fast.

Every module names the decision it serves. A module that serves no decision gets cut.

Step 5: Build in the hands-on exposure

Concept alone produces over-trust. For each tier, design at least one hands-on component where leaders work with a real model on something they recognize:

Watch a model handle, and then mishandle, a task from their own function.
Stress-test a vendor demo: ask it the question the sales deck avoided.
Walk a real decision they face through an AI-assisted analysis and interrogate where it could be wrong.

The point of the hands-on is not skill-building. It is calibration. They should leave knowing in their gut where the technology earns trust and where it has not.

Step 6: Choose format and cadence, tied to real decisions

Match the format to the calendar and the decisions:

Anchor sessions to live decisions from Step 1. A governance module lands harder the week before the board votes on a governance policy.
Prefer short, recurring, decision-attached sessions over a one-time offsite that fades. Literacy is a posture, not an event.
Sequence by dependency: capabilities and limits before investment judgment, because you cannot size a bet you do not understand.

Step 7: Define success as decision quality

Replace attendance metrics with decision metrics. For the program to count as working, name what observably changes:

The quality of the questions leaders ask in AI decisions (sharper, more specific, harder to bluff).
The decisions themselves: bets sized with opportunity cost named, governance gaps closed, vendor claims challenged before signing.
A before-and-after on a real decision the team faced, judged by the person named in Step 1.

If you cannot state how you would know the program changed a decision, the program is not done being designed.

Step 8: Output the program design

# Exec AI Literacy Program: (organization)

**Driver:** (what prompted this)
**Audience:** (tiers in the room)
**Maturity starting point:** (skeptical / experimenting / scaling / burned)

## Decisions this program serves
| Tier | Decisions they own | When they face them |
|---|---|---|

## Objectives (phrased as decisions)
| Tier | Objective (a capability, not a topic) |
|---|---|

## Modules
| Module | Tiers | Decision it serves | Hands-on? |
|---|---|---|---|

## Format and cadence
- Format: (offsite / recurring series / embedded in leadership meetings)
- Cadence: (sequence and timing, anchored to live decisions)
- Sequencing logic: (what must come before what, and why)

## Hands-on components
- (per tier: the model exposure that calibrates trust)

## Success measures (decision quality, not attendance)
- (observable change in questions asked)
- (observable change in decisions made)
- (the before/after on a real decision, and who judges it)

## Open questions
- (unresolved, including any  markers)

Step 9: Review