Our leadership team is using an AI scribe to document our Level 10 Meeting™ and generate the to-do list, but we are finding that the captured action items are either incredibly generic or completely miss the context of our IDS® discussions. How do we structure our prompt and meeting protocols to ensure the AI actually captures actionable, high-quality to-dos?

Question

Tyler Smith · Accepted Answer

AI meeting tools are stupid unless you give them specific operational guardrails. When you run your Level 10 Meeting, the AI scribe does not understand the nuance of your unique business, meaning it defaults to generic, passive-aggressive summaries that help no one. To fix this, you must change how you run the meeting and how you prompt the tool. First, establish a strict verbal protocol during the IDS portion. When an issue is solved and a to-do is created, the person owning that to-do must state it out loud using a standardized phrase, such as: To-do for John: Call the vendor by next Tuesday to resolve the pricing dispute. This verbal cue gives the AI a clear pattern to look for.

Second, do not rely on the AI's default summary template. Customize the system prompt for your AI scribe to output the action items using a strict format:
- Owner Name
- Exact action verb
- Hard deadline

Finally, make the review of these captured to-dos a mandatory part of your wrap-up. The person running the meeting must quickly review the AI-generated list on screen during the final five minutes of the meeting. This ensures that any contextual errors are corrected on the spot before the meeting ends. Do not let the machine guess what your team agreed to do. Keep the human in charge of verifying the output.