A client sent me a Zoom recording last month. Sixty-eight minutes. Seven executives. One agenda item that nobody finished.
I dropped the transcript into Claude. Asked it to score the meeting against the 5-Minute Leader Protocol the way a black belt examiner scores a kata. Cold. Unimpressed. Looking only at the form, not the effort.
The AI gave the meeting a 31 out of 100.
The CEO who ran it has a master’s degree, two exits behind him, and an EBITDA that would make most founders weep with envy. None of that mattered. The protocol does not care about your resume. It cares whether the sixty-eight minutes you just spent moved a decision, closed a loop, or assigned an owner. Most meetings do none of those three things.
If you want to score meetings with AI, this is the most useful exercise you will run all quarter. I want to walk you through how to do it, what the 5-Minute Leader Protocol actually grades on, and why the report will probably embarrass you the first time.
Why You Cannot Trust Your Own Memory About Meetings
You walk out of a meeting feeling like it went well. That feeling is a lie your brain tells you to protect the time you just spent.
Here is the problem. The human mind treats a meeting like a movie. You remember the opening scene, the loudest moment in the middle, and the ending. Everything else fades. If the ending felt good, the whole meeting felt good. If somebody made you laugh near the close, you walk out smiling and never notice that the actual decision got punted for the third time in a row.
This is why so many executives are surprised by their own calendar data. Microsoft’s 2025 Work Trend Index, which analyzed productivity signals from 31,000 knowledge workers across 31 countries, found that 60 percent of meetings are now ad hoc rather than scheduled, and workers are interrupted every two minutes during the traditional 9-to-5 workday. The leaders running those meetings still describe them as “productive.” The data describes them as a kitchen fire.
An AI transcript reviewer does not have your brain. It does not remember the joke at minute fifty-four. It does not soften the read because the CFO was finally engaged for once. It looks at the conversation as a flat document and grades it against a fixed rubric. That is the entire value. You need a referee who has not had breakfast with the team.
I have written before about how leaders mistake activity for execution and what to do about it. Meetings are where this delusion is most expensive. They look like work. They feel like work. And most of them are theater.
What the 5-Minute Leader Protocol Actually Grades
The 5-Minute Leader Protocol is a fixed checklist of behaviors that should appear in any decision-making meeting if leadership is actually present in the room. It comes from the 5-Minute Multiplier, our daily leadership product, and it exists because most leadership advice is too abstract to enforce. You cannot grade “be a better listener.” You can grade whether a decision was named, owned, and dated before the meeting ended.
Think of the protocol the way a chess coach thinks about opening theory. There are a small number of moves a serious player makes in the first ten turns. You do not need to know every line. You need to know the moves that distinguish a tournament player from somebody who learned chess from a cereal box. The protocol is the opening theory of executive meetings.
Here is what the AI is looking for when it scores a transcript.
First, a stated outcome. Within the first three minutes, did someone name what this meeting must produce? Not the topic. The output. “We are leaving here with a hiring decision on the VP role” is an outcome. “We need to talk about hiring” is a topic. Topics rot. Outcomes close.
Second, owner clarity. Every action item that surfaces in the conversation must be attached to a single human name before the meeting ends. Not a department. Not a team. A person. “Marketing will handle it” gets the meeting a zero on this dimension. The AI flags every action verb in the transcript and checks whether a named owner appears within two speaker turns.
Third, a decision verb. Somewhere in the transcript, words like “decided,” “approved,” “killed,” “shipping,” or “we are not doing this” must appear, attached to the original outcome. If the meeting ends without one of those verbs, the AI marks the meeting as a deferral. Deferrals are the most common meeting outcome in the world and the most expensive one nobody tracks.
Fourth, next-checkpoint specificity. A date. A trigger. A standing review. Vague phrases like “let’s circle back soon” or “we’ll keep an eye on it” score zero. The AI is looking for a calendar event with edges.
Fifth, airtime distribution. Did one voice dominate more than 40 percent of the talk time? In meetings I run, when a single executive blows past that threshold, the decision quality almost always drops. The AI counts words by speaker and flags imbalance.
There are a few more dimensions in the full rubric, but those five carry most of the weight. They are the radio tower. Everything else is a signal that bounces off them.
How to Actually Run This Audit (The Workflow)
You do not need a custom tool. You need a transcript and a prompt.
Most video conferencing platforms now generate a transcript automatically. Zoom, Teams, Google Meet, Otter, Fathom, Read AI. Pick one. Export the transcript as plain text. Strip the timestamps if you want a cleaner read, but the AI will work either way.
Then open Claude, ChatGPT, or any frontier model. Paste in the transcript. Ask it to score the meeting against the 5-Minute Leader Protocol on five dimensions: stated outcome, owner clarity, decision verb, next-checkpoint specificity, and airtime distribution. Ask for a score out of 100, a one-paragraph diagnosis, and the three highest-leverage changes for the next meeting.
The first time you do this, do not show the report to your team. This is for you. You are the variable that matters. The team is responding to the conditions you set.
Run it for five consecutive weekly meetings before you change anything. Five data points beats one. You will start to see patterns. The same meeting will score the same way every week, because the protocol is grading the structure, not the topic. If your Monday standup scores 41 four weeks in a row, the standup itself is the problem. Not the team. Not the agenda. The form.
Unlike traditional meeting feedback, which depends on a human who was in the room and therefore biased by what they wanted to hear, AI meeting analysis is structural. It does not know who paid for the lunch afterward. It does not know which executive is feared. It reads the conversation as text and grades the text. That is the unfair advantage.
If you want a more guided version of this work, the AI Mastermind runs cohorts of leaders through their own meeting transcripts together. The peer comparison is brutal and useful. Nobody learns faster than a CEO whose meeting just got outscored by another CEO’s meeting in front of seven peers.
What Your First Report Will Probably Say
I have run this audit with more than thirty founders now. The patterns are almost embarrassing in their consistency.
Almost every first report says the same three things. You did not state an outcome in the first three minutes. Half of your action items have no named owner. You used at least one variation of “let’s circle back” without a date. If your meeting was an engine, those three failures are the equivalent of running the motor with no oil, three loose spark plugs, and a piece of duct tape over the dashboard warning light. It moves. It will not move for long.
The second most common pattern is airtime. The most senior person in the room talks more than 50 percent of the time. They do not feel like they are dominating. They feel like they are clarifying, leading, providing context. But the transcript shows what the AI shows. A monologue with occasional nodding.
The third pattern is the most uncomfortable. The decisions that did get made were not the ones on the agenda. The team drifted into a side conversation that felt urgent and emerged with a verdict on something nobody had prepared for. The agenda item that was actually supposed to close in that meeting got pushed to next week. Again.
You were never disorganized. You were under-instrumented. The protocol gives you instruments.
Where This Stops Being a Productivity Hack
If you only use the 5-Minute Leader Protocol to fix your meetings, you will get back roughly four to six hours a week. That is the floor. The ceiling is something different.
When you start grading your own meetings against a fixed rubric, you stop being the person inside the meeting and start being the person above it. You become your own coach. The AI is just the chalkboard. You are doing the actual review.
That shift, from participant to observer of your own leadership, is the entire premise of the ADAPT Framework. Awareness, Direction, Action, Purpose, Transformation. Awareness comes first because nothing else moves until you can see yourself clearly. Most leaders cannot. Their meetings are the mirror they have been carefully avoiding for years.
I built Arcules to 150 employees with a team running on AI infrastructure that did not exist five years before we used it. I spent years inside boardrooms at Canon learning that the most powerful executives were not the ones who talked the most. They were the ones who heard the meeting before they entered it and reviewed it after they left. They had instruments. The rest of the room had vibes.
You have access to better instruments than any executive in history. You are not using them on the place where most of your leadership actually happens.
The Conversation You Are Avoiding
Here is the question I want to leave you with.
If you ran this audit on every meeting you led for the next thirty days, and the average score came back at 38, what would you do?
You already know the answer. You would not fire your team. You would not buy a new project management tool. You would build a different kind of meeting. The one where the protocol scores high because the form is right. The one where decisions close, owners get named, and nobody walks out wondering what happens next.
That is what leadership looks like when it stops being a feeling and starts being a discipline.
Run the audit this week. The transcript is already sitting in your account. The AI is already paid for. The only variable is whether you are willing to read the score.
