How we thinkArticle

AI is multiplying the demands on people and money faster than organizations can decide among them.

AI can strengthen evidence and accelerate execution. It cannot decide what the organization should value or who has the authority to commit it.

In this piece
  1. More demands, at the same capacity
  2. Beliefs move faster than the record can track
  3. Execution now outruns governance
  4. What AI can do here, and what it cannot
  5. What this asks of an organization

Most of what is written about AI and work is about doing things faster. We are more interested in what happens one step earlier, at the point where an organization decides what to do at all. That step has not become faster, and it is now under more pressure than it has ever been.

More demands, at the same capacity

AI lowers the cost of producing a plausible first version. A prototype, a market analysis, a draft plan or a working demo used to take weeks. Each now takes days or hours.

This sounds like pure gain, and for the person with the idea it is. For the organization it changes the shape of the problem. An idea with a working prototype behind it is a far stronger demand on people and money than an idea on a slide. When prototypes are cheap, many more ideas arrive in that stronger form. More ideas become credible demands on resources that have not grown to match.

The old filter was effort. Only the ideas someone cared enough to build a case for made it into the room. That filter is going away, and nothing has replaced it. So the list gets longer, every item on it looks more finished, and the same people have to choose among them in the same number of hours.

Beliefs move faster than the record can track

A priority rests on beliefs: that this work can be done, that it will have the effect we expect, that it costs about what we think. AI updates evidence and exposes new possibilities quickly. A belief that was sound when the plan was made can be out of date a month later.

Most planning runs on a slower clock. Priorities are set, written down, and revisited each quarter or each year. In between, the people doing the work learn things the plan does not know. They do the sensible thing and work to the new facts. The plan on file stays as it was.

The work may be right. The record is now wrong, and nobody chose that. Priorities set on last month’s beliefs need a visible way to be re-decided, not silently ignored. Without one, the organization ends up with two plans, the written one and the real one, and no account of how they came apart.

Execution now outruns governance

Whatever gets approved can be built and shipped fast. That used to be the slow part, and its slowness did some quiet work. A decision that took six months to carry out could be noticed, questioned and corrected along the way.

When carrying out a decision takes weeks, a poor one is finished before anyone has looked at it twice. Two teams can each build, quickly and well, things that contradict each other. Execution speed is starting to exceed most organizations’ ability to make, govern, and reconcile the decisions that are supposed to be steering it.

What AI can do here, and what it cannot

AI is useful in this problem as well as a cause of it. It can gather evidence, summarize what is known about an option, point out that two commitments need the same people, and notice when work in progress no longer matches what was decided. We use it for all of these.

What it cannot do is settle the two questions underneath every priority. The first is what the organization values, and how much. Whether this is a year for growth or for stability is a choice, and evidence can inform it without making it. The second is who has the standing to commit the organization. A ranking produced by a model that nobody with authority has adopted is a suggestion.

What this asks of an organization

None of this calls for a new method. It calls for doing a few ordinary things on purpose.

  • Decide what is eligible. When ideas are cheap, saying what kind of thing will be considered, and at what level, becomes the first act of prioritizing.
  • Keep the reasons with the decision. If the criteria and the evidence are on the record, a new fact changes one input and the order follows. Without them, every change reopens the whole argument.
  • Make re-deciding normal. Give a changed belief a place to go, so a plan can be changed openly between planning cycles.
  • Say who decides. Be explicit about who may commit the organization and who may change a commitment, so that speed does not turn into drift.

The organizations that do well with AI will not be the ones that produce the most. They will be the ones that can still say, at any point, what they decided and why.