The team sat through a ninety-minute AI session on a Tuesday afternoon. Everyone nodded. Everyone had a question. Everyone left the room saying “that was really useful.”
Monday arrived. Nobody could tell you what rule changed. Nobody could tell you who owns the next step. Nobody could tell you what gets tested first, or by when, or against which real work. The room ran clean. The energy was high. The follow-through was zero.
A keynote can leave people thinking differently. A workshop has to leave them doing differently.
Here’s the thing. Most of the AI training that gets called a workshop right now is a keynote with a Q&A tacked on the end. The leader plans a talk, invites the team, delivers slides, answers questions, and calls it a workshop because there were questions. There were also questions at the end of every keynote you have ever attended in your life. That is not what makes a room a workshop. What makes a room a workshop is that people leave holding an artifact they built together, and the business is different on Monday because they did.
The rooms are telling us what they need
I have spent the last few months in rooms full of leaders asking practical AI questions. The pattern is clean. A year ago the same rooms wanted inspiration. Right? “Show us what AI can do.” Now the same rooms want diagnosis and decisions. “Tell me what to change on Monday.” Inspiration created motion. The motion is now waiting on operating architecture.
You can hear the shift in the questions. What data is safe to put in a prompt. Whose account owns the tool if the employee leaves. Which of these three workflows deserves the first automation. Who reviews the AI output before it goes to a customer. Those are not curiosity questions. Those are operating questions. They come with a due date. And they will not be answered by a slide deck no matter how sharp the slides are.
The room I want to indict is not the room that asks curiosity questions. It is the room that asks operating questions and gets sent home with talking points.
Three room types, and the difference matters
Before you plan another AI session, name what kind of room you are actually building.
A keynote is a room where the job is to change what people see, believe, or notice. Best output: shared language, a new frame, energy, urgency. That is a real product. Every business needs it sometimes. The failure mode is pretending inspiration will convert itself into operating change while nobody is watching.
A workshop is a room where the job is to help people produce something usable together. Best output: a decision, a map, a rule, a scorecard, an owner, a next step, a working artifact. The dish is on the pass by the end. The failure mode is calling something a workshop because there were questions. Questions are not an artifact.
An advisory session is a room where the job is to diagnose one person’s or one company’s situation with expert guidance. Best output: a specific recommendation, a risk call, a narrowed decision. The failure mode is letting one attendee’s case take over a room that was promised a shared output.
All three are useful. None are interchangeable. AI application fails when leaders choose the wrong room shape for the outcome they need, and then blame the audience for not doing anything with it.
The habit is starting with slides
Here is the operator habit to indict. Leaders start with slides because slides feel like preparation.
You have felt this. A team meeting is coming up. You open a deck. You outline five sections. You write the first bullet. It looks like preparation. It feels like preparation. It is preparation. It is just preparation for the wrong outcome. The wrong preparation, done thoroughly, is still preparation for the wrong outcome. You end up with a beautiful deck for a room that needed an artifact.
The first planning question is not “what do I need to cover?” The first planning question is “what must this room leave holding?” And you cannot answer that with a topic. You have to answer it with a thing. A worksheet with the top three data types they identified. A one-page rule for who owns which AI tool. A ranked list of the four workflows they are going to try first. A three-column card with green, yellow, and red data by end of session. A named owner next to each item.
If you cannot name the artifact, you have not planned a workshop yet. You have planned a talk with breakout groups.
The kitchen is the room
Our Station Plan is the AI in Crayon framework for what an AI-native business looks like as a working kitchen. It has a Chef at the center holding taste, an orchestrator at the pass routing tickets, stations on the line producing the dishes, and a plated output that leaves the kitchen for the customer.
That same architecture runs a workshop. And it exposes exactly where most AI sessions fall apart.
The facilitator is the Chef. Not the smartest person in the room. The person holding taste for what a good outcome looks like. Their whole job is to protect the artifact and refuse to let the conversation drift into interesting-but-off-target territory. When someone in the room asks a great question that would take the room off the artifact, the Chef parks it, thanks the person, and pulls the room back to the plate.
The agenda is the Pass. Not a list of topics. A route. Each block of the session has a purpose, a handoff to the next block, and a specific piece of the artifact it is filling in. If a block does not hand something forward, it does not belong on the agenda.
The exercises are the stations on the line. Each one produces one thing. Not conversation. One thing. A list. A ranking. A rule. A next step. A scored decision. If the exercise produces conversation without a takeaway, the station made no dish.
The output is the plate. The decision the room made. The rule the room wrote. The owner the room named. The next step the room committed to. You can point to it. It has words on it. It has an owner’s name next to it. It leaves the room in somebody’s inbox before the parking lot clears.
The workshop fails when every station produces conversation and nothing ever gets plated. That is not a bad audience. That is a room with no kitchen.
Why AI adoption specifically needs this shape
AI adoption is not usually a motivation problem. Most operators are motivated. They read the launches. They saw the demos. They are tired of watching competitors talk about efficiency they are supposedly getting from tools those competitors also do not actually run yet.
AI adoption is a diagnosis and decision problem. People need to know which workflow is in bounds for this quarter. What the rule is when the AI answer looks wrong. Who owns the guardrail. What counts as progress in seven days. Which of the four candidate handoffs gets the first test slot.
None of those get answered by a talk. All of them get answered by an artifact the room built together with the Chef holding taste. If your team leaves your AI session without those answers on paper, they will go back to their desks, hit the same friction they hit yesterday, and quietly conclude that AI is not for them. Not because the tool failed. Because the room did.
Yeah. That is the whole diagnosis. The paid tools already exist. The training material already exists. What is missing is the room that turns the material into a decision your business can act on.
Does that make sense? The gap is not education. The gap is architecture.
A light callback to how we deploy the stations
One more callback, then the Monday Move. Our Professional Recipe has an ingredient called Output Over Process. It says the way you deploy an AI station is by defining what a finished dish looks like before you write the process to make it.
That ingredient is not just for AI stations. It is how you plan a workshop. Define the finished plate first. Then build the agenda backward from the plate. Every exercise has to fill in one line of the artifact. If you cannot map an exercise to a line on the artifact, cut it.
Same discipline. Different kitchen. The plate first. Then the recipe. Then the room. This ties directly to what we wrote about the governance card in Paid Tools Are Not a Governance Policy. The governance card is a workshop plate. A five-line artifact any room can produce in ninety minutes if the Chef stays honest about the outcome.
The Monday Move: write the plate before the deck
Before you plan any AI training on your calendar, sit down and write the one artifact the room must leave with. In one sentence. On one card.
Owner of the move: the person planning the session. Usually you, if you are the ops leader responsible for AI adoption. If someone else is planning the session, tell them the artifact you expect the room to produce, and let them design the agenda backward from it.
The artifact can be one of these. A decision. A rule. A scored ranking of candidate workflows. A three-color data classification card. A named owner for each of the top five AI tools in the building. A one-page handoff spec for the first workflow you are going to test. A green-light list of what your team can paste into an approved AI tool this week. Something. Anything with edges, a name, and one line at the top that says “by end of session, we will produce ___ .”
Guardrail: if you cannot name the artifact in one sentence, do not call it a workshop yet. Call it a keynote, an advisory session, or a planning meeting. Tell your team the truth about the format. A well-run keynote is worth more than a badly-run workshop. What you cannot do is deliver a keynote and hope the team turns it into an artifact after they get back to their desks. That has never happened in the history of Monday.
Visible check: at the end of the session, every person in the room can point to the artifact and say what changed. If they cannot, you did not run a workshop. You ran a talk with better lighting.
Next step: build the agenda backward from the artifact. Start at the plate. Ask what has to be true right before the plate leaves the pass. Ask what station produces that piece. Ask what the room needs to have decided before that station can run. Work all the way back to the opening. Every block has to earn its slot by filling in one line of the artifact.
So.
If the goal is to change what your team thinks about AI, give the keynote. Do it well. Own the format. Do not pretend it is more than it is.
If the goal is to change what your team does about AI on Monday, do not build a talk. Build the room system. Hold the plate. Route the pass. Run the stations. Ship the dish. Send everyone home with something in their hand that would not have existed without them.
A keynote can leave people thinking differently. A workshop has to leave them doing differently.
Write the plate. Then build the deck.
Original framework. Distilled from client work.
Framework spine: The Station Plan as workshop architecture, with The Professional Recipe’s Output Over Process as the design discipline for building the room backward from the plate. Read the Station Plan.
~ source material · Original framework. Distilled from client work.
