The leadership meeting on AI ran clean. Nine people at the table. Everyone had a question. Everyone had a good one. The Q4 marketing lead asked about a customer-service tool she saw at a conference. The head of ops raised a concern about who owns the ChatGPT account when the intern leaves in December. The founder wanted to talk about a browser-agent demo he had watched Saturday morning. Every question landed.
The meeting ended at 3:57. Everyone thanked the room. Everyone left with a small warm feeling of alignment.
Nobody could name what got decided.
By Wednesday morning, the customer-service tool question was still open. The intern’s ChatGPT access was still ambiguous. The browser-agent demo had disappeared from the follow-up email. The energy from Tuesday afternoon had literally not converted into a single change anyone could point to.
Here’s the thing. Nothing in that meeting was broken. The room was engaged. The questions were sharp. The founder had the right instinct to invite input on a topic he cannot see all sides of alone. The habit that ate the outcome is not a bad-meeting habit. It is a decisive-leader habit that quietly flips over when the topic is AI.
The habit is confusing inclusive with decisive. You invited the room in because you did not want to decide alone. Now you have a room, no decision, and a diluted next step. Participation was fuel. The room is not the steering wheel.
The habit is not “asking for input”
I want to be careful here because the habit sounds like it is asking the leader to stop asking for input. It is not. Rooms are useful. Rooms are also the reason most leaders got out of their own head on AI this year. Someone in the room caught a risk you did not see. Someone in the room raised the workflow you had not considered. Someone in the room told you the truth about a tool the vendor sold you last month.
The habit to indict is one step further downstream. It shows up in the sentence “let’s get everyone’s input on this” delivered at the front of a meeting where the leader has not yet named the decision the meeting is supposed to make.
Right? The order is the tell. When the decision is named first and the input is invited second, the room does its job. When the input is invited first and the decision is supposed to emerge from the input, the room drives to the softest place it can find. Consensus is not a decision. A room full of good input still needs one person to hold the outcome.
A team AI meeting that starts with “let’s talk about where we should be using AI” almost never ends with a decision. It ends with agreement about categories. Categories are not moves. You cannot deploy a category on Monday. Meanwhile the founder who called the meeting has quietly outsourced the decision to the room by not naming that a decision was going to happen. The room did not steal the wheel. The wheel was left on the table.
Deciding is not a passive stance
Our Chef’s Stance is the AI in Crayon lens on what the leader is actually doing across four postures. Doing, Dispatching, Dialing, Deciding. The whole point of the framework is that the Chef changes stance relative to one dish at a time. You are Doing on the recipe that has never been written. You are Dispatching on the recipe you just finished writing. You are Dialing on the recipe that is running but not clean yet. You are Deciding on the recipe that is running clean.
The reason it matters here is that most leaders read Deciding as a low-effort stance. You are stepping back. The recipe is running. The station is producing. You are just tasting the dish now, catching exceptions, deciding whether it stays on the menu.
That is exactly the stance that quietly evaporates in a room. Deciding is not passive. Deciding is the whole reason the room exists. If you as the Chef step out of Deciding when the room heats up, the room is not going to catch that. The room is going to keep serving up interesting questions. Nine sharp questions with no Chef will produce a room full of well-fed opinions and no dish.
The meeting is not deciding for you. The meeting is producing ingredients. Somebody at the head of the kitchen has to hold the outcome.
What the room is actually for
Here’s the thing. Rooms are useful because they can do things the leader cannot do alone. That is the reason they exist. Name what a room is for before you use it and the whole confusion goes away.
A room can surface risk the leader cannot see. The head of ops has watched three tools get sold to the sales team without ever hitting the security review. The room can raise that. The room can add texture the leader could not have generated alone. The Q4 marketing lead knows exactly which of five candidate workflows would break her calendar in November. The room can put that on the table. And the room can stress-test a decision the leader is quietly leaning toward. If four people raise the same concern, the leader learns something no other input path could have surfaced.
Those are the three real jobs of the room. Surface risk. Add texture. Stress-test. All three produce ingredients. None of them produce dishes. Ingredients are useful. Ingredients also are not the plate.
Yeah. That is the whole diagnosis. Rooms produce ingredients. Rooms do not produce dishes. When you treat the room as if the ingredients will assemble themselves into a plate by the end of the hour, the plate does not exist.
Room shape matters too. Whether you are running a keynote, a workshop, or an advisory session changes what the room is even capable of producing. If you have not yet named the room type before you sit down, start there. But even the best-shaped room still needs somebody at the head of the kitchen holding the plate.
Three lines at the top of the agenda
Here is the practical move. Before the next leadership meeting on AI, write three lines on an index card and pin it to the top of the agenda. Print it if you are meeting in person. Paste it at the top of the Google Doc if you are meeting on Zoom.
The decision that will get made in this room. One sentence. Concrete. “By 4pm today we will decide whether the customer-service AI pilot goes to the intake team or the escalations team.” Not “we will discuss the customer-service pilot.” Not “we will explore AI options.” A decision has edges. If your sentence does not have edges, the meeting will not have a decision.
The DRI who owns the decision. One name. The leader who called the meeting. This one is embarrassingly easy to skip. If you leave the DRI as “the team” or “leadership,” you have not yet decided who is deciding. And a decision without a DRI does not survive the walk to the parking lot.
The three inputs I need from the room before I decide. Three lines, no more. “I need to know the workflow volume for each option, the risk of a wrong customer response in each context, and whether either team has capacity to review the outputs for the first two weeks.” Notice what that does. It tells the room what it is here for. It gives the room a clear job. It also draws a boundary around what the room is not here to do, which is decide.
Then run the meeting to those three inputs. Every question that arrives gets sorted. Answer now if it feeds one of the three inputs. Park it if it is a good question at the wrong moment. Later section if it is already scheduled on a future agenda and would steal time from the current one. Every parked question gets an owner and a return point. Parked is not the same thing as ignored.
At the end of the meeting, write one sentence. “We decided X, the DRI is Y, the next step is Z, and it happens by W.” Send it to the room within the hour. If you cannot write that sentence when the meeting ends, the meeting did not produce a decision. And the next step is not going to survive Wednesday.
What this is not
One guardrail. This is not a case for railroading the room. It is not a case for pretending you already know the answer and running theater. Leaders who fake a decision are not making decisions. They are performing them. The room knows. The trust the room extends the next time you say “I want your input” is thinner every time you use it as a costume.
The move is the opposite. You name the decision because you actually intend to make it. You name the DRI because that person owns the outcome, and you owe the room a name so they know who to hold accountable. You name the three inputs because you are being honest about what the room can help with and what it cannot. Every one of those disciplines makes the room more useful, not less.
Same instinct. Different frame. You are still asking for input. You are just not handing the room the wheel.
So.
The next AI meeting you call, put the index card at the top. The decision, the DRI, the three inputs. Then run the room to those three inputs and hold the decision yourself.
Participation is fuel. The room is not the steering wheel.
Does that make sense?
Original framework. Distilled from client work.
Framework spine: The Four D’s (The Chef’s Stance), specifically the DECIDING stance. Read the Chef’s Stance.
~ source material · Original framework. Distilled from client work.
