Reading the Room

I want to write about a skill that I think is underrated by approximately ten orders of magnitude in the conversation about AI transformation. The skill is reading a room.

I do not mean reading a room in the polished, presentational sense — knowing how to land a joke at a board meeting. I mean a quieter, harder thing: sitting with a team for an hour and accurately understanding what they are feeling, what they are afraid of, what they are excited about, and what they are pretending.

The leaders who can do this are running AI deployments that look like magic. The leaders who can't are running deployments that are technically sound, organisationally well-funded, and failing. The difference is almost entirely on this one skill.

When you sit with a team during a period of change, there are several signals running in parallel and you have to learn to listen to all of them.

There is the stated content — what people are saying out loud. This is usually the least interesting signal. It is heavily filtered for what people think they're supposed to say to leadership.

There is the unstated content — what people are visibly not saying. The questions that never get asked. The objections that get raised privately afterwards but never in the room. The topics everyone steers around.

There is the energy — whether the room is leaning in or leaning back, whether the engagement is real or performative, whether the laughter is relaxed or anxious. These are not nothing. These are the most honest data in the room, and most leaders ignore them entirely because they're not on the slide deck.

There is the what people do when they think you're not watching. Whether they actually open the tool the next morning. Whether they ask the colleague next to them how it works. Whether they default back to the old workflow the second they're under time pressure. The room you really need to read is the room when you've left.

Every workplace change has an emotional dimension, but AI is unusual in two ways that make reading the room more important than ever.

First: the threat AI poses to identity is unusually direct. Previous waves of automation came for routine, repeatable work. AI comes for expertise. It comes for the thing your senior people built their careers around. It threatens the part of work that they thought was safest because it required judgment. The emotional impact of that is large, and it is largely invisible if you don't go looking for it.

Second: AI is so fluent and so confident-sounding that it triggers a specific kind of insecurity. Users are afraid of looking less competent than the tool. They will pretend to use it well long after they've stopped using it. Self-reports of AI adoption are wildly unreliable because the user knows what the right answer is and they say it. Only by reading the room, the actual room, can you find the truth.

I am not sure reading a room can be fully taught. But I am sure it can be developed, and I have watched leaders get noticeably better at it. Here is what they did.

They spent unstructured time with the team. Not all-hands. Not roundtables. The actual job, in the actual workspace, watching how the work is done. The body language of someone who is genuinely engaged versus performing engagement is unmistakable in person and invisible in a Zoom grid.

They asked open questions and shut up. Not "are you finding the tool useful?". That gets you a yes. Something like "walk me through what you did with it yesterday", and then nothing — silence — until the person fills the space with the real answer. Most leaders are physically incapable of staying silent that long. The ones who learn to are the ones who learn what's actually happening.

They tracked feelings the way they tracked metrics. Same rigour. Same cadence. They could tell you, with the same confidence they'd tell you about pipeline, whether morale on team B had dropped in the last two weeks and why. That kind of attention is unusual and it changes outcomes.

They acted on what they heard. This is the part that creates trust. If a team raises a concern and nothing changes, they stop raising concerns. Reading the room is only useful if the reading produces visible response.

I want to be honest about something. The cohort of leaders who can read a room well is, in my experience, smaller than the cohort that believes it can. There is a particular failure mode where a senior person becomes more confident in their EQ as their actual EQ erodes, because the people around them have learned to manage their reactions instead of giving them honest signal.

If you suspect this might be you — and almost no leader thinks it's them — the test is simple. Find the most candid person you trust. Ask them: in the last AI-related meeting we ran, what was being felt that I did not pick up? If they can tell you several things you missed, you have work to do. That work is the most valuable work you can do right now.

The technology layer of AI is going to get easier and cheaper every quarter. The human layer is going to get harder and more important. The leaders who win the next decade are the ones who treat reading their team accurately as the central skill of their job. Not a soft skill. Not a complement to the real work. The actual work.