AI is changing the way we work, and most of us have figured that out by now. What gets discussed much less is that it’s not changing it the same way for everyone: whether you’re a manager or an individual contributor, the same thing hits you in very different ways and places. Even though, looking far enough ahead, the endpoint might be the same for everybody: ending up a manager.
If you’re a manager
If you’re a manager, delegating is part of your job. With AI you find yourself with what is effectively an army of cheap doers that on many tasks perform on par with (on just as many worse, on some better than) flesh-and-blood collaborators. And in the world of remote teams – where the people you work with are participants in video calls, names on emails and profile pictures in Slack messages – the difference is less and less noticeable. Not just in the quality of what you get back, but in the day-to-day experience itself. There’s a misconception I see plenty of managers fall into, though: thinking this stuff is mostly about other people. We always talk about AI as an accelerator of personal productivity, so in execution terms: those who produce get impacted, those who decide orchestrate. Sounds comfortable, if you think you’re sitting on the right side of the desk.
But if we think about what most management roles are, at their core, we notice they rarely involve big strategic decisions. Much more often it’s a continuous stream of small ones: setting a priority, approving or postponing, assigning a task to the right person, deciding whether something is good enough to ship. Decisions that, more than anything, require context. And with well-organized context, the best models are often already able to make them, those decisions.
What remains harder to hand over to a machine, for now, is not so much the decision: it’s putting your name on it, answering for what happens. So, all in all: I’m not at all sure the impact will be stronger on execution than on management. What’s certain is that a lot will change for everyone.
If you’re an individual contributor
If instead you’re not a manager (maybe because you don’t want to be, fair enough), AI greatly increases the amount of things you can do. And as a consequence your job changes quite a bit.
If your role used to be effectively – even if maybe not exclusively – productive and executional, things now get complicated. People who prefer this kind of role often do so precisely to avoid all the stress and the chores that come with managerial positions.
And there’s a bit of a dilemma. If you don’t use AI, you’re certainly less productive and therefore less valued inside your organization. Maybe called out by management, maybe cut out – depending on how AI-pilled the organization you work in is. If you use it heavily instead, the amount of stuff you produce grows a lot, and your ability to make sure that what gets produced is actually up to what you’d want it to be shrinks. Maybe, in some ways, so does your ability to feel that work as “yours” (especially when it doesn’t come out well).
So you find yourself in an objectively uncomfortable position: on one side you pretty much have to use these tools to keep up, on the other using them can expose you to responsibilities you partly feel aren’t yours. Because you would have done that thing differently. Or you wouldn’t have done it at all, if we were all still in the old world.
The friction
A good manager is used to taking responsibility for work produced by others too. They accept a series of trade-offs on the degree of control and quality they can get, and in exchange they focus on creating the conditions for others – at this point I’d say humans or agents, indifferently – to do their best work.
When you’re a contributor, you may simply not have that vocation. And that’s exactly the point: a friction gets created where you either accept being dramatically more productive, losing at least part of the control and consequently taking on responsibility for things you didn’t, in a way, generate yourself, or you risk not making full use of these technologies and suffering consequences of a different kind.
Now, granted that there are many known ways to get a certain degree of control even over work generated by others – after all, it’s what managers have always done – it remains true that for organizations, and for the people in them, the very fact that this friction exists is objectively a risk.
So much for the present. But this story has another piece: what happens when processes really start running on their own, and who puts their name on things in place of the systems.
The job of taking responsibility
One thing, for sure, automation doesn’t eliminate. Quite the opposite. A process can even run on its own, but the responsibility for what it produces… no: someone who answers for what happens has to be there anyway. For customers, for the law, for shareholders, for the rest of the organization.
And that’s why, as processes become automatable, people willing to take responsibility on behalf of these systems will become more and more valuable. The executional work is done by the machine; what stays scarce – and therefore, for now, precious – is someone putting their name on it.
This is nothing new. In many industries there are already figures who are there mainly for this: the driver of a vehicle that could drive itself is the first example that comes to mind. They’re not there for their driving skills; we keep them because we need someone to hold accountable if something goes wrong. In organizations it will be more or less the same: the head of this or that department who simply has to be there, because someone has to take the responsibility.
Then again, I don’t think this will hold forever. When organizations become truly autonomous, and these systems have earned a trust they don’t have today, the very need for these figures – and the very concept of responsibility – will change once more. But that point is relatively far away, and in between there’s a long phase where taking responsibility will be an increasingly demanded job and an increasingly rewarded disposition.
All managers
Taking responsibility for work you didn’t execute directly: which is exactly part of the manager’s job. And that’s why, in the end, individual contributors as we know them might stop existing, and we’ll all become managers: of humans or of agents.
In a world like that, I believe many contributors will find the step easier than you’d think – at least inside properly designed organizations – because many of the reasons that have historically kept them away from those roles will fade. If the workforce is made of AI instead of humans, the politics and the relational side of things will change to the point of – maybe – making these roles less daunting for those who have avoided them all their lives. Managing ten, a hundred, a thousand agents doesn’t take the same energy, and the same mediations, as managing ten, a hundred, a thousand people.
And in a world where not many organizations let contributors climb the hierarchy the way managers do, this could lead to a reshuffling of the managerial picture in many companies: certain skills that have been decisive until today will count less, and new profiles will be able to climb organizations more easily.
It’s also true that managers start with an advantage of a different kind, having always been used to these practices: delegating, accepting trade-offs on control, creating the conditions for others to produce. Which happen to be exactly the things it takes to work with agents. A race it is not — but if it were, it wouldn’t be decided in either direction.
Conclusions (for now) So what?
Organizations need to be designed so that people can use these tools unlocking their potential, creating a healthy environment that also accounts for a new concept of “responsibility” that follows from a new way of working, staying aware of the quantity-quality trade-off that exists today. Those who do it first gain an advantage; those who don’t push people to use AI in secret, or in the widespread mode of slowly, otherwise I’ll get everything done way too fast.
Managers need to understand that their job is shifting even more toward what has always been its core: creating the conditions for others – humans or agents – to produce well, and taking responsibility for the result. Knowing that a part of what they do today, made of small contextual decisions, is as automatable as executional work, and then some. And that without solid knowledge of the domain they operate in, they’re less and less relevant.
Individual contributors are facing a change of perspective more than a change of duties: value is shifting from doing to getting things done, and paradoxically the new world might open doors for them that the old one kept shut. For those who have dodged management their whole lives, managing agents might be the least painful way of becoming one.
Then again, as always, six months from now I might want to rewrite this post entirely. But this looks like the direction to me.