A founder recently asked me for advice. His company is pushing an AI transformation and got stuck on how to allocate token quotas across employees, which is the direct cost of computing power. If they distribute it equally, top performers run out of budget. If they allocate differentially, management cannot provide a credible rationale.

The question was never about token quotas. It was about the baseline quality of his management. Strategy can be articulated grandly, and vision statements can sound inspiring. But the moment you slice the cake, where the knife falls, who gets more, and who gets less is a real money transaction. Everyone sees it with complete clarity.

Resource allocation is the only management action that cannot be faked or blurred.

Whether a company can allocate resources and capital well directly reflects whether it has competent management. Only well-managed companies know where resources and capital should be concentrated. When everyday operations are run vaguely, management has no idea where to place the knife when allocation time arrives. Equal distribution and arbitrary guesses become convenient shelters to avoid making real business judgments.

Most companies fail at allocation because they remain blind to four fundamental realities in daily operations.

First, they do not know who their true top performers are. Status reports make everyone look exceptionally busy. Once you strip away the polished corporate presentations, management has no clear idea who is actually driving the business forward.

Second, they do not know what the right performance metrics are. Many companies mistake working hours for contribution and process compliance for business outcomes. Scorecards and evaluation forms are filled out completely, while the vital operational signals that determine commercial survival are ignored.

Third, they do not know what employees actually do all day. As long as someone sits at a desk typing on a keyboard, management assumes output is being generated. The amount of truly effective, value-producing time is so small that the employees themselves would panic if they stopped to calculate it.

Fourth, they do not know which specific tasks actually create value. Extensive workflows exist solely to give multiple layers of management peace of mind. People exhaust their energy on internal status memos, pouring their best effort into tasks farthest from paying customers.

These sound like four separate problems, but they are four facets of the same reality: management does not know where business value is actually generated.

None of these are new problems created by AI. In the past, nobody forced you to provide an exact answer. Setting token quotas now simply forces the issue into the open all at once.

Once these four realities are clear, quota allocation ceases to be an issue. Top performers leverage the tools to expand business volume, and their legitimate quota is virtually unlimited.

AI merely shines a spotlight on the truth. It exposes that management never knew who to terminate in the first place, or where their remaining workforce should focus their effort.

Faced with this confusion, the instinctive reaction of many operators is to sit in an office and design an increasingly complex allocation spreadsheet. That is using administrative bureaucracy to mask operational ignorance.

The less clarity leaders have about underlying operations, the more convoluted their spreadsheet weightings and conversion formulas become. In the end, this produces dedicated administrative headcount just to fill out data sheets, solves zero problems, and completely ties the hands of frontline operators. The damage goes further. As spreadsheets grow more intricate, management attention shifts from running the business to debating allocation formulas. Eventually, the entire organization serves the division of the pie instead of baking it.

Good management relies neither on ad hoc improvisation nor on moral appeals. It must be anchored in operating mechanisms. Open the throttle directly and let API usage records and commercial output speak for themselves. Operational logs are the new general ledger.

When an operator uses tools to generate incremental business, do not restrict their resources. Pursue them with capital and support. For employees who cannot figure out how to leverage modern tools, the issue was never whether their quota was too high or too low. The real question is whether that seat should exist at all.

Capable managers are inherently scarce. Among the capable ones, middle managers tend to excel at identifying top individual performers and observing daily operational workflows. Senior executives tend to excel at calibrating performance metrics and defining value-creating tasks. Very few leaders excel at all four.

In a well-managed company, resource allocation is simply the natural byproduct of business performance. Every cut of the knife is computed directly by the numbers, with zero room for guessing. The popular observation that the world is run by amateurs holds truth, but a serious operator cannot stop at cynical resignation.

If you cannot allocate resources clearly, it is simply because you have been running a sloppy war.

Based on personal operating experience. Cases are anonymized.