The universal professional AI skill
Meeting follow-up is one of the most universal professional workflows. It exists in HR, finance, accounting, operations, product, sales, and compliance. It is also a strong AI use case because the input is natural language and the output needs structure.
The risk is subtle. AI can turn vague discussion into fake certainty. A person says "maybe Alex can look at this next week," and the model writes "Alex owns this by Friday." That is not productivity. That is invented accountability. The skill above prevents that by requiring confidence levels and marking implied actions as needing confirmation.
This is also a good example of how professional AI adoption is changing. Tools such as ChatGPT Work and Microsoft 365 Copilot are designed to operate inside everyday work context. But the value still depends on the workflow definition. A meeting skill should know what a decision is, what an action item is, what evidence is missing, and what a manager must confirm before sending.
| Output | AI can prepare | Human must confirm |
|---|---|---|
| Decisions | Extract explicit decisions from notes. | Whether the decision is final and correctly worded. |
| Actions | Draft owner/date/dependency table. | Owner, due date, and scope when ambiguous. |
| Risks | Surface blockers mentioned in the meeting. | Priority and escalation path. |
| Follow-up | Draft message and next steps. | Commitments before sending. |