AI Is Now Your Team's Loudest Clock

AI Is Now Your Team's Loudest Clock.

It has made individual production abundant. Collective judgement is becoming the constraint.

Individual work got faster across the team. Drafts, analyses and options arrive sooner than they used to. And yet something about how the team works feels slightly off.

The team is starting to feel less like a team and more like a set of fast individuals in adjacent lanes. People produce more on their own, then meet to reconcile what they each produced. The meetings that used to be where the team thought together have quietly become where it stitches parallel work back into one piece. Nobody decided this. It just happened.

Two earlier pieces in this series traced the same fault line at different altitudes. One followed the manager holding the gap between fast execution and slow decision-making. The other followed the individual absorbing that mismatch inside their own head. A later piece will follow it up to the board, where strategy is still sensed and decided on a slower clock than the world now moves to. Each is about work passing between people, or between an organisation and its market, at speeds that no longer agree.

This piece sits inside the team. Teams have a property that individuals and org charts do not: they fall into a shared rhythm, and that rhythm decides what kind of work they are still able to do together.

Teams keep time to something

A team does not simply work at a pace. It keeps time to something. There is almost always a dominant signal setting the beat: the leader's level of urgency, the delivery tool and its notifications, the sprint boundary, the quarterly cycle, the most anxious person in the room.

That signal is rarely chosen on purpose. A team locks onto whatever is loudest, most powerful or most immediate, and organises itself around it without ever discussing the decision. The trouble is that the loudest signal is seldom the one that matters most. A team can keep perfect time to its internal meeting cadence while drifting badly out of time with its customers. The rhythm feels coherent from the inside and is wrong from the outside.

It is worth being clear about what good looks like, because the answer is not a single, uniform tempo. Healthy teams run several rhythms at once: open periods for exploration, heads-down periods for individual production, clear points where they converge on a decision, and enough stability that people can anticipate one another. The problem is not that a team has more than one speed. It is that a team collapses onto one loud beat without choosing it, and loses the others.

AI is the loudest signal, but it does not set the clock by itself

Into that picture arrives the most insistent signal most teams have ever had. By making individual output close to instant, AI produces work at a speed nothing else in the room can match. A first draft, a model, a set of options, a rewrite, available in the time it takes to ask. Against that, every human rhythm sounds slow.

But AI does not set the team's clock on its own. A tool cannot impose a tempo. What happens is more ordinary and more avoidable: once near-instant production is possible, expectations reorganise around it. Faster output becomes faster expected turnaround, which becomes more frequent review, which becomes a standing assumption that the team should move at the speed of the thing producing the work. The technical clock becomes a social clock. That conversion is a choice, even when no one experiences making it, which is precisely why it can be made differently.

The asymmetry: generation got faster, judgement did not

Here is the mechanism underneath the discomfort, and it is sharper than "everything sped up."

AI accelerates some kinds of team work far more than others. Used in the obvious way, it makes generation cheap: options, variants, drafts and plausible directions, produced before anyone has decided what deserves attention. It does much less for interpretation, judgement and commitment, because that work is reciprocal. It happens between people, not inside a tool.

Follow what that does to a team. When each person can produce more finished-looking work alone, work enters the collective process later, and in a more complete and more defended form. People arrive with artefacts rather than forming thoughts together. The team generates more divergence and does less joint exploration. Meetings stop being where ideas take shape and become where finished pieces are reconciled and approved. The output rises. The shared understanding underneath it thins.

This is the part to hold onto. More divergence needs more convergence: more of the work that pulls a widening set of options back into one course of action the team actually owns. Yet convergence is exactly the slow, interactional work the fast clock is squeezing. The team is asked to converge more at the moment it is practising convergence less. That is the convergence deficit, and it is why faster production does not automatically make a team more effective. It can make a team busier and less collectively intelligent at the same time.

The general form of this is familiar from the rest of the series. Speeding up one activity does not remove the slower ones it depends on. It raises the pressure at the interface between them. Here the fast activity is individual generation, and the interface is the team's capacity to make sense of the output together.

When polish arrives too early

The same shift quietly changes who shapes the work.

When everything arrives looking finished, the moment at which an idea enters the team moves later. It is harder to bring a half-formed thought into a room where everyone else has turned up with something polished, and the half-formed thought is precisely what collective thinking runs on. A junior contributor's output can now read as assured as a senior's, which makes it harder to see whose judgement is actually carrying a decision.

The interesting effect is not that AI flattens status. It relocates it. Influence moves away from visible authorship and towards the quieter acts that now matter more: who framed the question, who chose which options to pursue, who decided what was good enough. Those acts happen early, before the polished artefact exists, and often without anyone noticing where the real judgement was exercised. A team can look more egalitarian on the surface while concentrating its actual discretion in fewer, less visible hands. The earlier in the work that ideas have to look finished, the less of the team is genuinely in the room when the direction is set.

Why the same tool helps one team and strains another

What I keep noticing is that the same AI pace lands completely differently in two otherwise similar groups.

One team absorbs the new speed and uses the time it frees to do more of the slow, shared work, not less. The other transmits the same pressure straight through, and faster individual output becomes always-on expectation, fragmented attention and a low hum of urgency that never resolves. Same tool, same pace, opposite effect.

The team's existing rhythm appears to be one important reason why. A team that had protected its convergence time, with clear boundaries and trusted ways of disagreeing, has somewhere to put the new speed. A team already organised around interruption and urgency simply acquires a louder clock. It is not the only factor: how tightly the work is coupled, the quality of decision rights, the level of psychological safety, staffing and the leader's own behaviour all move the outcome too. But the inherited rhythm is the one most often mistaken for something else, usually the resilience or calibre of the individuals, when it is really a property of how the team was already keeping time.

Design the rhythm rather than inherit it

None of this argues for slowing down, and it is not a defence of meetings. Plenty of collective work is wasteful, and AI can genuinely improve the good kind: preparing better, surfacing assumptions, drawing in quieter voices, cutting the time spent simply exchanging information. The scarce resource is not time spent together. It is high-quality collective judgement, and the aim is to protect that, not to protect slowness for its own sake.

Doing so is a design task, not a matter of discipline. It means separating the work of generating options from the work of converging on a decision, rather than letting the first crowd out the second. It means deciding, deliberately, at what point AI-assisted work should enter the group, so that ideas are still soft enough to be shaped together rather than arriving finished and defended. And it means setting explicit points of convergence and protecting them, holding down how many options are allowed to proliferate before the team commits, and asking whether more output is producing faster decisions or merely a longer queue to review.

This is what I have been calling Requisite Pace, applied one level down: a team is viable not when every activity runs at the same speed, but when its linked activities run at speeds that work together, and when it can move between them on purpose. The loud clock makes that harder by pulling everything towards a single tempo. Designing for it means keeping the clocks distinct.

The work you do together is now the scarce thing

AI has made individual production abundant. It has not made collective judgement abundant. In many teams it has done the opposite, multiplying the material that has to be interpreted, challenged and brought back into one course of action, while leaving the team's capacity to do that roughly where it was.

So the question is not how fast the team can go. Individual speed is no longer the constraint. Look at your own team's week instead. Where are ideas still formed together, rather than reconciled after the fact? Where does divergence stop and convergence begin, and has anyone left enough time for the second? Which parts of the team's cadence should follow the tool, and which have to stay human?

We do not yet know what sustained exposure to the loudest clock we have ever built does to the way a team thinks together. But we already know enough not to let it set the beat by accident. AI may now be the loudest clock in the team. It should not automatically be the one the team keeps time to.