Governance
AI guardrails
The limits I set when AI touches my leadership work. They describe my own workflow and are not legal or policy advice for your agency.
| What I let AI do | What I don’t |
|---|---|
| Draft leadership replies | Any send or public post |
| Flag scheduling conflicts | Client or PHI fields |
| Summarize public research | Clinical advice claims |
| Log decisions (scrubbed) | Staff or student identifiers |
My rule: I approve every send. No client data goes to AI.