The Room Is the Prompt

This is not a post about your agency.

At least, not directly.

It is about the agency you create, or unintentionally suppress, in other people.

If you are reading this, you are probably someone’s context: a manager, teacher, parent, founder, team lead, or just the person whose reaction determines what people are willing to say next. You are not only an AI user. You are part of the room someone else is responding to.

That is why I have not been able to shake something Daniel Pink said on the podcast.

We were talking about agency (the willingness to act without being told, to notice what needs doing and take the first step) when Dan said this:

“Agency is a function of context.”

I loved it immediately. Then I realized I did not just love it. I needed it.

Because I have spent years telling people to be more agentic. Look for opportunities. Keep the monkey on your back. Do not wait for someone else to hand you the assignment. Be the prompt.

More recently, I wrote that the bug is the brief: in the age of AI, the premium is moving upstream from problem solving to problem finding. Notice the friction. Name the annoyance. Commission yourself to act.

I still believe every word of that. But Dan’s line complicated (it in a useful way!).

Agency is not just a trait. It is not something you either have or do not have, like eye color or an unfortunate relationship with cilantro. It is also a response to an environment.

Dan flipped the camera around. He pointed it at the room.

I should have seen this coming

Years ago, Laszlo Bock told me something from his time leading People Operations at Google that lodged in my brain. The way I remember him explaining it, one of his biggest insights from Google’s obsessive data driven approach to people management was “performance is a function of context.”

The implication was wild. If you have an underperformer, do not immediately assume the person is the problem. Move them. Change the manager, the team, the project, the conditions. Often, their performance reverts toward the mean. In plain English: it improves.

Even wilder? If you have a star performer, be very careful moving them. Their performance can revert toward the mean too. In plain English: it declines.

Same person. Different context. Different result.

I have believed that about performance for years. I have used it to challenge lazy stories about “A-players,” because talent does not float around in the abstract. It shows up in a particular role, on a particular team, under particular constraints, with particular incentives, in a particular season of life. Change the context and you may change the person you get.

I knew that about performance. I had not fully applied it to agency.

But once Dan said it, I could not unsee it. Some rooms ask people to notice, choose, risk, try, and own. Other rooms ask people to wait, comply, protect themselves, and make sure they are not the one holding the bag when something goes wrong.

Then we act surprised when the second room produces low-agency people. (I know, I know. Mysterious.)

The inconvenient question

That is the inconvenient part.

You manage people. Or teach them. Or raise them. Or sit in meetings where your reaction determines what other people are willing to say next. You are not only an agent inside your own environment. You are part of the environment someone else is responding to.

That changes the question.

For years, I have heard some version of this from leaders, teachers, and parents: Why are they so passive? Why do they wait to be told? Why will nobody take initiative? Why do I have to push everything?

Those are not bad questions. Sometimes they are fair. Some people really do need to take more responsibility.

But they are also convenient questions, because they locate the whole problem inside the other person. Which means they quietly absolve the room. And, inconveniently, the person who built the room. (Hi.)

Dan’s line asks a more uncomfortable question: What is this environment teaching them to do?

Because environments teach constantly. A weekly staff meeting teaches. A classroom teaches. A performance review teaches. A family dinner table teaches. The way a manager responds to a half-formed idea teaches. The way a teacher grades an unexpected answer teaches. The way a parent reacts when a kid makes a mess trying something on their own teaches.

The official message may be, “We want initiative,” while the room is saying, “Do not move until someone important says it is safe.” The official message may be, “We value creativity,” while the room is saying, “Bring polished ideas only.” The official message may be, “Think like an owner,” while the room is saying, “Run it by me first.”

And the room usually wins.

The room before the prompt

This is why Dan’s line matters so much in the AI era.

In The Bug Is the Brief, I argued that most AI advice starts one step too late. It teaches people how to write better prompts. Useful, yes. But prompts are downstream of briefs, and briefs are downstream of noticing. The prompt before the prompt is noticing what deserves attention.

But there is a room before the prompt too.

People do not notice in a vacuum. Or more precisely, they may notice, but they learn what to do with what they notice. If every complaint gets labeled negative, people stop naming friction. If every rough attempt gets evaluated like a final product, people stop showing drafts. If every small experiment requires three approvals, people stop experimenting. If every person who volunteers gets handed an unbounded pile of extra work, people stop volunteering.

The bug may write the brief. But context determines whether anyone feels allowed to read it.

That is the part AI exposes.

For years, a low-agency room could hide behind the cost of execution. Of course people waited for assignments. Of course work needed approval. Of course someone had to define the project carefully. Doing anything meaningful required time, budget, coordination, and permission.

But now the first proof is cheaper. The memo, the mockup, the analysis, the prototype, the customer response, the tiny bot that solves the annoying thing your spouse has been asking about for twenty years — all of that is more reachable now.

Which means the constraint is easier to see. It is not always capability. Sometimes it is permission. Sometimes it is fear. Sometimes it is the learned habit of waiting.

AI does not just reward high-agency people. It exposes low-agency rooms.

Run the initiative replay

Here is the audit I would run. Not five pillars. Not a maturity model. Please do not make this a maturity model. (Someone is already opening Keynote. I can feel it.)

Just run the replay.

Think of the last three times someone around you took initiative without being asked. What actually happened?

Maybe they got thanked. Maybe they got silence. Maybe they got corrected for skipping a step or told, “Next time, loop me in first.” Maybe their rough prototype got treated like a final deliverable. Maybe their small experiment quietly became a permanent responsibility.

Whatever happened was the lesson — not just for them, but for everyone watching.

And if you cannot think of three examples, that is also the finding. It may mean nobody is taking initiative, or it may mean people are taking initiative quietly, where the room cannot interfere. Neither is great.

Make initiative cheap to attempt

I am not suggesting we abandon standards.

Some initiatives aren’t great. Some experiments need constraints. Some ideas should not ship. Some people confuse “high agency” with “I did a thing nobody asked for and now everyone must applaud.”

No. Agency is not license.

But if every attempt has to arrive polished, permissioned, and politically safe, do not be surprised when people wait. High-agency rooms make initiative cheap to attempt without making it consequence-free.

They ask better first questions:

• Not “Who approved this?” but “What did you notice?”

• Not “Is this finished?” but “What is the smallest proof?”

• Not “Why did you do it that way?” but “What did you learn?”

• Not “Can AI do this?” but “What would be worth trying with AI?”

Those questions change the room. They do not eliminate accountability; they move accountability to the right place. The standard is no longer “Did you wait correctly?” It becomes, “Did you notice carefully, try responsibly, and learn in a way the room can use?”

That is a very different room.

Before you audit the people, audit the room

I still want people to be more agentic. I still want students to look for opportunities, employees to find problems worth solving, executives to stop outsourcing curiosity to the youngest person in the room, and my daughters to develop the reflex to say, “I can take a first pass.”

I still believe the individual has a job. Context matters, but context is not destiny. Even in a low-agency room, you can notice the bug, claim the brief, make the first proof, carry the monkey for twenty-four hours, and see what happens.

But Dan’s line added the missing half.

If you are designing the room, you do not get to complain that nobody in the room takes initiative until you have asked what your room is prompting them to do.

If AI responds to prompts, people respond to rooms. Some rooms prompt waiting; others prompt ownership. Some rooms prompt self-protection; others prompt curiosity. Some rooms prompt polished compliance; others prompt small, useful attempts.

So the next time you catch yourself asking, “Why are they so low agency?” try turning the camera around. Ask what the room makes safe to notice. Ask what happens when someone claims a brief. Ask whether a small proof can travel here.

Because if agency is a function of context, every leader, teacher, and parent is a context designer. The people around us are always answering the prompt our environment gives them.

So yes, keep telling people to take initiative. But before you audit the people, audit the room.

Because the room is the prompt. And the question is, “What is your room asking for?”

Related: The Bug Is the Brief
Related: Be the Prompt
Related: Take Your Own Job Before Someone Else Does

Join over 38,147 creators & leaders who read Methods of the Masters each week.