Back to the blog

/en/blog · writing

Corporation or artel: what happens to software development when the floor of programmers is no longer needed

For decades software development was built like a corporation with floors of juniors. AI is taking over the lower floors, and an artel becomes the better shape: a master, a senior and one apprentice. Companies, education and the masters themselves have not adjusted yet.

Authors: slavb18
  • AI
  • Management
  • Hiring
  • Education

The head of a mid-sized IT company with twenty years in the trade admits it is easier for him to spend an extra day working with AI than to hire ten junior developers and deal with their burnout. He now works three times as much because of it, and he does not enjoy it.

You hear this kind of admission in many companies now. A task like wiring a couple of webhooks between systems, which a year ago meant hiring a fresh graduate, is done by one experienced person with AI in a few hours. People already live by the new rules, while companies, universities and the job market are still built around the old ones.

A corporation with floors

Software development was built like a corporation. At the bottom sit many juniors doing repetitive work. Above them are team leads, above the team leads are managers, and off to the side is HR with team-building events, grades and retention programs.

Education served the same pyramid. Mass training produced the average specialist: someone tired of their previous job took a course, became a tester and moved on to a stream of similar development tasks. Each of us knows dozens of such stories, and ten years ago it was a sensible path.

The pyramid paid off as long as repetitive work was cheaper to give to people than to automate. Now AI takes exactly the lower and middle levels of skill. It strengthens the top level: an experienced person who understands how the process really works gets a speedup from AI, even though it takes more time to set the task and check the result.

What replaces it

What replaces it is a form older than the corporation: the artel, a small cooperative of craftsmen. There is a master you can trust with a whole line of work, a senior next to them, and one apprentice who learns by doing. Fifteen juniors under the senior are no longer needed; AI does their work under the master's supervision.

The leader's attention gets concentrated in this setup. It used to be spread over twenty-five people: who is ill, who is down, who wants a raise. Now it goes to two people, and the bond with each has to be an order of magnitude deeper than the whole sum of relationships with the old department. If one of the two leaves or burns out, an entire line of work stops.

What trust rests on changes too. In a corporation trust is professional: there is a grade, certificates, a review process, and that is enough to work with a stranger. In an artel that does not work. You trust a person with a whole line of work only when you share the same idea of what is right and what is wrong. Without shared ethics all that is left is a one-off contract, and it is simpler to do the work yourself.

Hiring changes accordingly. An apprentice is taken on based on whether you will get along, and certificates say almost nothing. Whatever skills are missing, they will learn from the master and from the same AI tools. The only way to check this is to work together for a while; a test or an interview will not tell you.

Nobody has really adjusted yet

Companies see the benefit and cut costs. The temptation is easy to understand: make the three best people do everyone else's work and stop hiring for the lower floors. As a result, experienced people work three to five times more, and juniors have nowhere to learn.

Education keeps producing specialists for a lower floor that has almost disappeared. Courses promise a new profession in six months, even though this is exactly the step AI closes first. Corporate training programs do not help either: a company trains for its current tasks and does not know which skills it will need in five years.

The masters are not ready for the new role either. For twenty years they grew up in a pyramid where leading meant handing out tasks and watching deadlines. In an artel leading means raising an apprentice and keeping a deep bond with two or three people. A team-building event will not replace that.

And nobody has answered the main question yet: where future masters will come from. A master grows out of an apprentice who spent years getting their hands on simple tasks. If AI does the simple tasks and nobody takes on apprentices, in ten years there will be no one to become the senior. The "master and one apprentice" model keeps continuity only if apprentices are actually taken on and taught.

What an artel does not solve

An artel grows slowly. Its capacity is limited by the number of masters and their attention, and masters cannot be hired; they are grown over years. An artel takes on a large order by joining forces with other artels, and then the hardest part becomes agreeing on the boundaries of work and shared rules between them. The industry has no ready way to do that yet.

It is fragile. One person leaving a department of twenty-five is barely noticed, while in an artel of three it knocks out a third.

It asks more of the leader. Decisions that in a corporation spread across committees and approvals are made by one person in an artel, and the responsibility is theirs too.

And it leaves the questions we used to hand to HR: how to keep a person, how to come to terms with them, how to get through a conflict. In an artel these questions become personal.

What to do about it

Ask yourself honestly how many people the work really needs. If there is someone you can hand a whole line of work to, they have no use for a department under them. They need one apprentice and tools.

Take on apprentices even when it looks cheaper not to. It is the only way to have someone to pass the work on to in ten years.

Choose people by shared ethics and check it by working together. Skills go out of date in a couple of years, while ideas about what is right barely change.

Corporations for writing code will not disappear tomorrow. But the floor of programmers they stood on is already emptying, and the winners will be those who rebuild as an artel before the market forces them to.

Back to the blog