We use an AI tool to summarize our Level 10 Meeting to-dos, but the output is often filled with passive language like 'the team will look into this' instead of clear accountability, which is leading to confusion after the meeting. How do we train our AI workflow to generate crisp, EOS-compliant to-dos?

Question

Tyler Smith · Accepted Answer

AI meeting assistants are incredibly powerful for capturing notes, but they lack the operational context of the Accountability Chart. When an AI summarizes a discussion, it defaults to polite, passive language that completely dilutes the absolute ownership required by the EOS® process.

To fix this, you must train your AI tools with a highly specific system prompt. Feed the AI a strict set of rules for generating to-dos. Instruct the tool that every single to-do must be assigned to exactly one individual, never a group or the team.

Next, dictate that every action item must begin with a physical action verb, such as email, call, draft, or schedule. The AI must also be instructed to exclude any conversational fluff, vague updates, or background context. The final output should be a clean, bulleted list containing only the owner's name, the action verb, and the specific deliverable.

Finally, the Integrator must remain the ultimate filter. Before you exit the Level 10 Meeting™, spend the last five minutes reviewing the to-do list live on screen. Read each to-do aloud as it was captured, and make any necessary corrections on the spot. AI is a great assistant, but it cannot replace human ownership and operational discipline.