We want to use an AI tool to automatically summarize our weekly Level 10 Meetings and generate our To-Do lists, but the resulting summaries are always overly long, unstructured, and fail to capture the true intent of our decisions. How do we constrain AI to generate clean, EOS-compliant outputs?

Question

Tyler Smith · Accepted Answer

AI can be an incredibly powerful tool to run your operations, but only if you give it strict operational boundaries. When you let an AI meeting assistant run without constraints, it behaves like an administrative assistant who does not understand business operations. It dumps pages of useless transcripts and vague action items that nobody will ever read.

To make your AI tool EOS-compliant, you must feed it a highly specific prompt template that reflects the Level 10 Meeting structure. Stop asking for general summaries. Instead, program your AI tool to output only three specific elements.

First, have it list the exact weekly To-Dos generated during the meeting, formatted strictly as who, does what, by when. If the AI cannot identify a clear owner and a seven-day deadline, it must omit the item entirely.

Second, instruct the AI to generate a list of issues that were identified but postponed, so they can be easily moved to your long-term issues list on the V/TO.

Third, require a bulleted list of any decisions made during the IDS portion, limited to one sentence per decision.

Never allow the AI to draft paragraphs of narrative or try to capture the emotional context of the debate. You only need the hard outputs, the commitments made, and the next steps. By enforcing these strict structural limits, you eliminate administrative waste and ensure your weekly operational data remains clean, actionable, and ready for your next pulse.