When the future of work comes up, the usual question is which jobs disappear; but that question does not help today's decision. Change rarely happens at the level of a job. It happens at the level of a task. Every job is a bundle of dozens of tasks, and what shifts over the next few years is the share of tasks a machine can take.

The practical consequence is clear: jobs are recomposed far more often than they are eliminated. An employee who spent most of the day gathering and tidying data keeps the same title while the content of the day changes. A manager who sees this stops asking which roles vanish and starts redefining what value each role is meant to add.

Tasks whose share shrinks

• Collecting and moving data: extracting from one system and entering into another, or producing a recurring report.
• The first draft: writing the initial version of a text, some code, a design or meeting notes.
• Answering repeat questions: whatever is already in the documentation and only needs finding.

What becomes more expensive

• Framing the problem correctly: the machine is strong at answering and weak at noticing that the question itself is wrong.
• Judgement under ambiguity: decisions where the data is incomplete and responsibility for the consequence rests with a person.
• Trust and relationships: negotiation, handling an unhappy customer, aligning people; areas where tool productivity means little.
• Assessing output quality: whoever can tell that a fluent, confident answer is wrong is worth more than whoever gets answers faster.

The manager's role changes rather than disappears

Part of a middle manager's work has been relaying and consolidating information, which is precisely the most automatable part. What remains is heavier: setting priorities, owning a decision the system recommended, and holding together a team whose working order new tools have disturbed. Organizations that miss this remove the middle layer and then discover what they gave up.

The Iranian labour market reality

Two opposing forces act at once in Iran. On one side, the emigration of skilled staff has made senior experience hard to reach, which turns tools that raise a small team's productivity from a choice into a necessity. On the other, young educated workers are plentiful and the relative cost of human labour weakens the incentive to automate. The result is a slower but unavoidable transition; a company that postpones upskilling ends up buying both tools and new people a few years later.

A practical example

Consider an organization whose finance team spent much of its time producing monthly reports. After data collection was automated, nobody was let go and no hours were cut; the same team began answering questions it had never had time for, such as which product's margin was quietly thinning. The difference was that the manager had decided in advance where the freed hours would go.

A 90-day starting path

• First month: in one team, list the tasks and estimate the time share of each; that list reveals what the role really is.
• Second month: pick one frequent, low-risk task and, before any tool, write down what the freed hours will be spent on.
• Third month: review the outcome with the team and update the role description; a role not updated on paper does not change in the employee's mind either.

Common mistakes

• Acquiring a tool before deciding where the freed hours go; the result is more productivity in work of no value.
• Presenting automation as a headcount-reduction plan; the team then withholds data and genuine cooperation.
• One-off generic training instead of practice inside daily work.
• Eliminating entry-level roles; if juniors get no hands-on experience, there are no seniors in later years.

Frequently asked questions

• Which skill should be taught first?
  The ability to evaluate output; using a tool is learned in hours, recognizing a wrong answer is not.
• What about a team that resists?
  Resistance is usually fear of replacement; it eases when the destination of the freed hours is clear, not when a memo is issued.
• Should we hire new people with AI skills?
  In most cases upgrading the existing team, which already knows the business, produces results faster than external hiring.

Takeaway

By the end of this decade, the difference between organizations will owe less to their tools and more to how quickly they recompose roles. Preparing is not hard: start with one team and one task, and write down where the freed hours go before you automate anything.

Glossary

• Upskilling: raising the skills of current staff for new responsibilities.
• Reskilling: preparing a person for a role different from their current one.
• Hybrid role: a role where a tool performs part of the work and judgement stays with the person.
• Human in the loop: keeping the final decision with an accountable person, even when execution is automated.
• AI literacy: enough understanding of a tool's capability and limits to use it properly.