For leaders adopting AI
Use AI to widen judgment without outsourcing responsibility.
AI can surface alternatives, compress analysis, and draft a persuasive answer. It cannot decide which obligation matters most, which cost is honorable, or what kind of institution the decision creates.
The decision beneath the decision
The practical AI governance question is not only what the system may do. It is what the leader must still own. When the answer is uncomfortable, speed and fluency can make abdication feel like sophistication.
The Five Modes expose different AI risks. Holding can become rigid refusal. Restraining can become passivity. Eroding can rationalize endless compromise. Growing can become grandiose transformation. Embedding can become control disguised as standardization. A code gives each strength a boundary.
How the work proceeds
Diagnose the pattern. Work the decision. Write the code.
- 01
Begin with the free Mode Finder to identify the leader’s pressure pattern and shadow.
- 02
Map the decision points where AI can analyze, recommend, draft, or act, and where a named human must sign.
- 03
Write the line, restraint, test, review loop, and transmission standard for the organization’s real AI use cases.
Typical outputs
Questions before a conversation
What buyers usually need to know.
Is this technical AI governance?
It complements technical, legal, security, and model-risk work. Its focus is executive judgment and accountable decision ownership.
Can we begin without an organization engagement?
Yes. Start with the free Mode Finder and the AI judgment resource. A team workshop is the next step when the boundary must be shared.
What use cases fit the workshop?
Consequential uses involving people, credit, strategy, reputation, allocation, or institutional commitments are the strongest candidates.
Illustrative workshop scenario
When an AI recommendation becomes a decision by default
An executive team receives a confident AI-generated recommendation but cannot trace several of its claims. The session maps where AI can assist analysis and where a named person must verify evidence and accept responsibility.
The decision record
Specify the permitted inputs, evidence checks, human decision owner and escalation conditions. Test what happens when the model is plausible, confident and wrong.
A sample working agenda
- Before the session: agree the decision, attendees and handling of confidential material.
- Establish the facts, assumptions, decision authority and deadline.
- Use the team’s self-reported Mode Finder patterns to question habitual responses.
- Compare alternatives, name the cost and write the commitments.
- Assign the first action, review date and conditions for reconsideration.
This fictional example illustrates the work. The agenda and retained documents are agreed when an engagement is scoped.
Chris brings experience leading B:Side Capital and B:Side Fund and teaching entrepreneurship. Read his B:Side CEO letter and view his ASU teaching profile.
Prepare for the conversation
Inquiry
Discuss the decision your team is carrying
A short description is enough. Chris will review it directly before the conversation.