Asher's Blog

For years, companies could explain average work by pointing to limited capacity. There were not enough people, the right specialist was too expensive, the research would take too long, the data was difficult to analyse or the team did not have the technical skill to produce what management wanted. Some of those explanations were genuine, particularly for growing businesses operating with limited capital and talent.

Artificial intelligence does not make every constraint disappear, but it makes many capabilities dramatically cheaper and more accessible. A small team can now research a market, analyse customer conversations, draft campaign variations, inspect financial patterns, document a process, prototype a product and write basic software faster than it could a few years ago. The work still needs supervision, but the cost of producing a credible first version has fallen.

This is why AI may not immediately replace your staff, but it will replace many of the excuses that once protected weak execution. When capability becomes cheaper, the bottleneck moves. The question is no longer only whether the company has enough resources. It is whether its people possess the judgment and will to use the resources already available.

Capability is becoming abundant

Businesses used to organise themselves around scarce expertise. A report required an analyst, a design required a designer, software required an engineer and a campaign required a copywriter. These professions remain valuable, but intelligent tools increasingly allow one capable person to cross the boundaries between them.

This does not mean the work of a specialist has become worthless. It means the specialist can operate at a higher level while routine production becomes easier. An accountant can spend less time formatting reports and more time interpreting what the numbers imply. A marketer can generate several drafts quickly and spend more attention deciding which message deserves to represent the company. An engineer can accelerate repetitive coding and focus on architecture, security and product consequences.

The economic consequence is significant. Companies that adopt these capabilities intelligently should be able to produce more value with the same number of people. Companies that refuse will not necessarily lose every employee to a machine; they may simply become too slow and expensive compared with competitors whose people use machines well.

The new bottleneck is judgment

AI can produce an answer, but it does not automatically know which answer matters. It can generate a strategy that sounds coherent without understanding the company’s actual constraints, customer relationships or moral obligations. It can identify patterns in the data while missing that the data itself is incomplete or biased.

Judgment is the ability to choose among plausible answers. It asks whether the recommendation fits the objective, whether the evidence is sufficient, which risk deserves attention and what consequence will follow if the system is wrong.

This is particularly important in a business like payroll or financial services. A generated message may sound reassuring, but somebody must decide whether the company has enough evidence to make the promise. A model may recommend approving a transaction or extending credit, but a responsible organisation must understand the data, controls and customer consequences surrounding that decision.

As AI improves, weak judgment becomes more expensive because people can now produce sophisticated mistakes at greater speed. The output looks better, arrives faster and can spread throughout the organisation before anyone has asked whether it is true.

Will matters because tools do not create urgency

Many companies already know what should be improved. Their customer complaints are documented, employees have proposed solutions and competitors have demonstrated what is possible. The obstacle is not information. It is the willingness to decide, change priorities and follow the work through.

AI cannot give a leader the will to confront a poor performer, stop an unproductive project, rebuild a weak process or admit that a cherished idea is not working. It can organise the evidence, model alternatives and draft the difficult message, but somebody still has to accept responsibility for the decision.

This is why tool access will not produce equal results. Two companies can use the same model and obtain very different outcomes. One asks shallow questions, accepts the first response and adds “AI-powered” to its marketing. The other improves its data, interrogates the output, tests it against reality and converts the learning into a repeatable operating system.

The advantage belongs less to the company with access and more to the company with disciplined intent.

AI exposes unclear work

Employees often struggle to use AI because the task itself has never been clearly defined. If the company cannot explain the objective, the relevant context, the expected standard and the constraints, no prompt will rescue the work.

This exposure is useful. It reveals where managers have been assigning activity rather than outcomes. “Prepare a marketing plan” is vague. “Design a three-month experiment that produces a defined number of qualified opportunities within this budget, using these customer insights and reporting these measures” gives both the employee and the tool something meaningful to work with.

The quality of AI adoption will therefore depend partly on management quality. Leaders must become better at defining problems, documenting context and describing what good looks like. The tool does not eliminate management; it punishes ambiguous management more quickly.

Do not use AI to demand permanent overwork

If AI makes people faster, some leaders will conclude that employees should simply carry more work. This may create short-term output but waste the deeper opportunity.

The objective should not be to fill every minute saved with another task. It should be to improve the quality, scale and economics of the organisation. Some capacity should move towards better thinking, customer understanding, experimentation and learning. Otherwise, the company automates production while preserving the same weak decisions.

Leaders should also be honest about how productivity gains will affect roles. Telling employees that AI will change nothing while quietly using it to reduce headcount destroys trust. The company should explain which work will disappear, which responsibilities will grow and how people can develop the judgment required for the new environment.

The standard has moved

Average work will become more difficult to defend because the tools required to improve it are increasingly available. A poorly structured document, an unexamined customer pattern or a repetitive manual process can no longer be excused as easily by saying the team lacked capacity.

The fair question is not whether every company can afford the most advanced system. It is whether leaders are actively learning what is now possible and applying suitable tools to their most important constraints.

AI will not remove the need for capable staff. It will increase the value of people who can define problems, evaluate evidence, exercise taste, communicate with humanity and take responsibility for outcomes. It will also expose people who have confused activity with value.

Capability is becoming cheaper. Judgment remains scarce, and will remains personal. The companies that win will not be those that merely possess artificial intelligence, but those whose people have enough intelligence and determination to use it well.