HealthPartners Leader Says Start AI With Meeting Agendas. Not Exactly Rocket Science. That Is The Point.
Graves, vice president of home and community services and value-based strategies at HealthPartners in Minnesota, told medical group administrators to stop treating AI as a sweeping technology initiative. Her advice: know your organization's AI policies, approved tools, and governance structure first. Then experiment with trivial tasks like drafting a meeting agenda, updating a job description, or building a new-hire orientation checklist.
This illustrates the principle of constrained onboarding, or what I would call minimum viable experimentation. The mechanism is simple. Start with low-stakes tasks where failure costs nothing and success builds organizational confidence. AI adoption fails when organizations treat it as a transformation project rather than a daily habit. Graves gets this right, even if the advice seems almost embarrassingly obvious.
Graves at HealthPartners in Minnesota, where she leads population health, care management, home-based care operations, and value-based strategies. She delivered this guidance through MGMA, which I assume stands for something those of you in healthcare administration would know.
- Write down three repetitive administrative tasks you do weekly, like drafting agendas or updating checklists. Pick the most boring one. That is your first AI candidate.
- Open whatever AI tool your organization has approved. If none are approved, ask. If no one knows, that is your first problem to solve.
- Ask the AI to draft your chosen task. Example prompt: 'Draft a one-page new-hire orientation checklist for a medical practice receptionist.' Review the output. Edit it. Use it. Repeat tomorrow with a different task.