Hello everyone! Today I want to stop and take apart one very topical and painful question. For many years, more than 20 years, I have been working in people management, and now more than ever I hear people's fear about their own relevance: will AI replace me, will I be left without a job, what career opportunities will I have at all. I have a personal observation. Sometimes the specialists who are more afraid are the ones who use AI more deeply and more often, because, in my view, they understand all the capabilities of the tool much better. But that is only my hypothesis, so I wanted to look at the forecasts and the real figures in general. Today I want to take the US market as an example. I take the US not because it is the whole world or because its figures can somehow be transferred to the markets of Asia, Russia or Europe, but purely because changes around AI show up there quickly and there are plenty of fresh sources, some released even in the first week of September, that we can rely on. Today is September 2026, and in a quarter or half a year these figures may well change, some trends may change. Today I want to suggest working out where we are now in terms of labour markets,
whether we see any mass layoffs connected with AI and around AI, as much as people talk about it now and as the headlines scream. To work out what is happening with hiring, what vacancies are appearing, what skills stand behind those vacancies, and what new career opportunities all of us can discover and see for ourselves. So, we will ask each figure different questions. First of all, who counted it, over what period, on which market, a percentage of what number — because, as always, the devil is in the details.
The first scary figure. Why is it scary? Because it sounds quite frightening and is very easy to put into a screaming headline. Since the beginning of the year we have seen 116,175 announced job cuts connected with AI. That figure is taken from the Challenger, Gray and Christmas report. It is an American company that for decades has been tracking public statements, public announcements by employers about job cuts and hiring plans. It is an important source — not official statistics, but essentially an important source that is well regarded in labour markets.
So, Challenger records the reason the way the employer itself formulated it. And for January to August 2026 employers named AI as the reason in 116,175 announced job cuts. That is about 21% of all the job cuts inside the Challenger database for the period. That is, 20% of all the job cuts recorded by the research company for the period from January to August 2026 are cuts connected with AI in one way or another. In fact, there is a lot of detail inside that. If we go inside — what does a job cut connected with AI actually mean?
It can be connected with AI, and with restructuring, and with cost cutting, and so on. But all the same, such a figure exists, and you can meet it quite often in various sources. And the overall market in that same database looks a lot calmer, because if you take all the announced job cuts for the 8 months of 2026, that is 41% fewer than in 2025 over the same period.
If you exclude the public sector from that sample, it is 15% fewer. Sometimes such details are very important so that we understand the real dynamics, the real movement. So I think it really makes sense to include such comments, so that we all understand. Technology companies need to be looked at separately. Here the picture is much more tense, since Challenger was talking about the whole commercial market. If we talk about technology companies, then since the beginning of 2026 they have announced 155,126 job cuts for all reasons.
That is 52% more than a year earlier, and roughly 29% of all the job cuts in the Challenger database. But the technology market is the employer's industry, not a person's profession, and a technology company employs not only developers but also HR, marketers, finance people, sales and lawyers. So job cuts in tech do not mean, and are not always equal to, cuts of developers alone. A logical question arises: how much of the total number of job cuts in the technology market fell on cuts connected specifically with AI?
What percentage is that? Unfortunately, we cannot say for certain, because I did not find exactly that figure and exactly that breakdown, so I would rather leave that gap than invent percentages. But let's look at what this figure — 116,175 job cuts because of AI in 2026 on the US market — means overall, and how large it is for the whole economy. To get a feel for the scale, I compared this figure of announced cuts over 8 months with the 159,075,000 jobs in the American employer payroll statistics for August.
And it comes to roughly 0.07 percent, or about 7 positions for every 10,000 jobs. And this is not the probability of being laid off and not the share of people who really lost their jobs. We are dividing an 8-month flow of announcements by the stock of jobs at one moment in time. It is just a rough ruler that I wanted to use so that we understand and assess how large this figure is — or perhaps how small.
I do not mean to say that the fear is made up because of this: after all, if you are inside that zone, then for you it is not statistics, for you it is a real event. But this big figure, one way or another, does not describe the risk for each person. And before being frightened by a scary headline, before being frightened by some trend that is very seductive today and by all this information noise, it is always better to work it out and understand what you are relying on. My next observation, from studying and diving into various figures, is that the market seems to have frozen a little.
And what led me to that observation? There is the US Bureau of Labor Statistics, a federal statistical agency that tracks all vacancies, actual hires, voluntary quits and separations initiated by the employer. There is no AI reason there, but what is interesting is that in July 2026 there were 7,271,000 open vacancies. That is 2.6% more than a year earlier, but actual hires were only a little over 5 million.
So this is, of course, not the conversion of any one vacancy, but the combination looks like caution on the part of employers. Why do I say the market seems to have frozen: vacancies are opening, hiring plans are growing, but it is as if nobody plans to hire. They hold back a little, look for the right person, watch what will happen tomorrow, the day after, in a month on the market, and adjust the opportunities they offer on the labour market to a wide range of specialists. The next point in favour of the market having frozen is that employees have started leaving on their own less often.
And this is very interesting analytics, because that same bureau tracks precisely people leaving to change jobs, for another role, another position, another company. So, there have been fewer such moves, which says that people have started relying more on their current employer, taking a wait-and-see position, wanting to make decisions properly in this crisis — I would even say not crisis, but rather uncertain — period of time. Separations initiated by the employer are 6% fewer than a year earlier.
So my interpretation of all these figures is this: the market seems to have frozen a little, employers are holding vacancies and hiring more cautiously, and, I repeat, people are leaving on their own a bit less often. And a general wave of layoffs — the so-called layoffs around AI, or not connected with AI — we are not seeing at the moment. So I do connect part of this caution with AI, with a lack of understanding of the future model of work, but at the same time that is my own personal observation.
I think there are, of course, colleagues who share it and those who do not. Your opinion is interesting here — what do you think? Share it in the comments, and I am sure we will find plenty of reasons to reflect on these topics in our next episodes.
So what does AI actually do to headcount inside a company, what happens there? There is probably one more interesting source here — a working paper by the Center for Economic Studies at the US Census Bureau. That research unit surveyed businesses and used the answers of more than 117,000 unique companies. And the data was collected from November 2025 to February 2026. And among the companies that used AI in their work, 95.7% reported that over the previous 6 months AI had not changed their total headcount.
2.3% reported a rise in headcount and 2% a decline. So the conclusion for the question we posed for today's podcast is this. Across a broad mass of companies, AI so far much more often changes the work inside the organisation. We will talk about that in a moment. And that, it seems to me, is a very strong link to all our previous episodes and to all the topics on our channel in general, rather than changing the total number of people. That does not contradict Challenger. Challenger still sees the large public announcements, while the survey of companies shows a more everyday picture.
But these are, so to speak, different views of reality, of what is happening on the market today, how much to be afraid, where to look, where to run. Today's market, and the influence of AI, cannot be extended to each person's loss of a job. Just as it cannot be extended to each person's gaining a job. I think in general there are different levels to rely on if we are talking and want to appeal to such figures.
One level is the employer's industry. And right at the start I gave an example of how much more the current changes affect the technology market. The second is the person's profession, because even inside the technology market professions differ very much. The third is the specific task inside the profession. And obviously there are tasks today that, potentially — let's call it that — we can replace more. And there are tasks that, in general, we probably cannot or could not replace.
So before we say that AI is destroying all jobs, let's say which industry we are talking about, which profession and which task.
Now I suggest moving on to the second part of our story. We have taken apart and talked through, in reality, some footholds we found for what is happening on the market in terms of that hiring and firing, and talked a little about how the market has frozen and about that fear that AI will replace everyone. But I also want to look here at what new things the market is starting to buy. Because within the technology contour it is somewhat easier to quickly formulate groups, some new groups.
Today we see vacancies on the market for those who create AI — for example, the position of AI engineer. We see vacancies for people, for engineers, who embed AI into products and processes; there may be so-called AI strategists, or there is a vacancy today that is growing very strongly. That is Forward Deployed Engineer — an engineer for the applied deployment of AI, who works with the client, understands the business problem, connects the model to a real process and brings it to a working result.
Within the technology contour there are also specialists needed who ensure the reliable operation of systems, who carry these changes through the company. So it would seem these new positions arising in the AI era are somewhat more obvious. But the most interesting part for me begins not only inside the technology professions: I looked at real vacancies today and analysed them in marketing, sales, finance, HR, recruiting, support and administrative work. Everywhere there the same pattern repeats, and I want to talk about it now.
The company is looking not only for a person who performs some function with the help of AI; it is looking for a person who improves that way of working. I want to go through several positions right here. In marketing, for example, they are looking for a Marketing AI Operations Manager — a head of marketing AI operations. Not a person who simply makes content: they have to own the whole technology system of marketing and, on top of that, manage internal AI agents. Another vacancy in marketing is agentic marketer — I would translate it as a marketer who builds and uses AI agents, again.
There candidates are directly asked to show a working marketing agent, for example one built on n8n and Claude Code. And alongside that, the requirements still keep 4-plus years of experience in digital or growth marketing, SEO, an understanding and knowledge of how funnels are built, conversion, campaign strategy, client communication and so on. In sales a new title appears — go-to-market engineer, an engineer of commercial processes and of the system for going to market, for example combined with automation.
That is, a company is looking, for example, for a head of automation of the go-to-market system. That is, a person who owns the CRM as a product and automatically assembles a map of companies, people and signals, and creates AI agents for finding clients, enriching data and growing accounts. Or there are quite a lot of vacancies in sales in general today. I found such go-to-market engineers — specialists who connect Salesforce and other systems, automate account scoring, bolt on AI to understand and assess all the intent signals, to assess and analyse the work and effectiveness of sales-department specialists, and so on.
But in parallel with that, these people are still expected to have 5–8 years in revenue operations, working with revenue, working in sales operations, the ability to work with SQL and BI, to understand the funnel, to have commercial planning skills, and so on. So such a role at the intersection becomes very strong. In finance, in particular, even Apple — well, not 'even', Apple is just one of my examples — Apple is looking for a Product Manager, AI Finance Transformation, that is, a manager of an AI product for transforming finance.
Together with finance leaders they must see the real business pain, redesign the process, choose the right AI scenario, translate it into requirements and, of course, define what data the AI agent receives, how explainable the result is, and how to measure accuracy and effect. There is a lot in financial planning — specifically, I found vacancies connected with AI transformation analyst. In HR and recruiting the same thing is happening. Today there are vacancies for Automation and AI HR Transformation — a product owner for AI automation in HR transformation.
Such a person, again, defines certain behaviour for AI agents, restrictions, points of mandatory human checking in the process. They are looking for Recruiting Operations and AI Automation — heads of recruiting operations and AI automation, a person who owns and understands what an Applicant Tracking System is, who embeds AI into sourcing, embeds AI into communication with candidates. In the same way, of course, especially in an area as sensitive to the use of AI as HR and recruiting, such a person understands the points of human control and human management inside the process.
In support and administrative work, standard requests can be automated more and more often, and what is highly valued is working with exceptions, understanding confidentiality, knowing when automation must be stopped. So, in all these functions — I am not even retelling the companies I went through — in all these functions I see what seems to be one and the same formula.
That is, essentially: do the work — the work has to be done — and rebuild the way it is done. That is, a new request appears, and it is not at all about AI taking a vacancy or AI taking a job; instead, new layers of the professional profile are being formed. And in the end I arrived at several pillars, so to speak, that we can rely on. This is purely my interpretation, and it may or may not resonate. Share what you think about it. How do you see this change happening specifically in your professional field, because I listed a very limited number of professions.
So, the first layer, which is the basic one and has not disappeared from a single set of requirements, is domain expertise. You need to know your function so well that you understand where that AI can be bolted on at all, or not. Whether something can be automated or not. That deep, detailed understanding of your area, the ability to form professional judgements correctly, the ability to see deeply and perhaps anticipate and highlight all the risks. An understanding of how all the details in your specific domain are interconnected.
The second layer — let's call them layers — is process architecture. From a deep understanding of the domain and expertise it follows that a specialist must be able to break their work down very clearly into processes, into components, into tasks. To understand, in each process, what result they want to get, what trigger starts the process, what stages the process has, who takes part, who is responsible, what tools are used, what deadlines are inside, what exceptions there are.
That is, to break their work down into details and into separate tasks. And that is possible only with a deep understanding and with depth of one's professional judgement in general. The third layer is to assemble some working solution on the basis of the first and second. What else struck me in all the examples I gave earlier is that, essentially, each specialist is also asked to be a bit of an engineer. There are very few places where there is that contact and the ability to formulate a task for some technical specialist.
That is, the very task of working with AI, of automating one's own processes, becomes a requirement precisely for that deep domain specialist. That is, each of us becomes a little bit of an engineer. And any strong specialist today must be able to assemble that small prototype, or be able to describe the logic to another engineer so precisely, with all the constraints and results, as to be responsible not just for some idea, for what they see can be done, but for the result.
The next layer — moving on — is about quality. Always about quality, because some... We must always understand — we have talked about this a lot on our channel — regarding quality, and with any... With any bolting-on of AI, with any automation, we must always understand what result we got, what we earned, what we saved, what we achieved, how much better our process became, how we understand at all where a human check is needed, and when to stop the system, and which errors are acceptable.
Well, this is a very multi-faceted, multi-component layer. And the last layer that I single out for myself in all the vacancies, in all today's requests, is implementation — yes, the implementation layer — because you can be a very strong specialist, you can understand and cut the elephant into pieces, you can be able to assemble some engineering solution precisely and understand its result, but it still has to be carried through the business, through information security, through the lawyers today.
This, it seems to me, is a really big topic in general — the safety of artificial intelligence in businesses. You have to train the team, you have to work through the cognitive resistance, which, you know, seems not to disappear — it just moves, over time, during this era, from one block to another. But the most important thing is that the base of this whole chain is professional expertise and an understanding of the process. In essence, even the technical part — that third layer and the engineering solutions — can very often be assembled, and you can find out what to assemble, by asking AI itself.
That is where a certain paradox lies. But that professional depth — domain depth, professional depth and an understanding of what you do and how your work is built — is what lets you ask the right question in order to find the right solution. So the way I see it, we do not need some identical list of courses for everyone.
And if we simply answer the question of what to invest in now and how to work with this fear that AI will replace us, then it seems to me extremely important now to deepen your profession, whatever industry you work in and whoever you work as, to learn to describe processes, at least to understand their components, to assemble small AI solutions yourself, to try on some small part, on the most understandable process. Perhaps to describe just one working process, assemble that AI solution on it and see what changed, what data we got, and try to implement that process inside your company, inside your own business perhaps, inside some personal project of yours.
That is, essentially, you can start not with a new position, not with some ten tools, but with exactly what you do as part of your work every day. I think it is certainly noticed — and it seems to me we have touched on this many times, repeatedly, one way or another — that this requirement of at least basic AI literacy, an understanding of your tasks, of the tasks where you can use AI to strengthen yourself, to strengthen the quality of your work,
is already becoming a default requirement, as knowledge of Microsoft Office once was, say, on a CV. But that is the first part of the answer to the question. Obviously, yes. The second, as we have said more than once, is that there is in general a shift in whom a particular business is looking for and how well a person's previous experience matches this new vision — the one we talked about today. And there is definitely a mismatch between how much the employer's requirements and what the business is looking for have changed, and what the person offers, because very often they rely on their previous experience, very often on their previous knowledge, in which they were successful and in demand, and they do not always see those areas for their own development, including with the use of artificial intelligence.
While companies have already moved ahead, they already understand that they absolutely must close new areas. So this is an answer for a separate podcast, because there are a great many components built into it. We need to work out whom companies are actually looking for, and exactly which people, with which competences, in which professions cannot find work. That, you know, Olya, is a good example. I also came across it, by the way, while searching for various figures now. There are positions that are not deliberately cut, but people leave and they are not reopened, because businesses understand internally that these tasks have become very easy to redistribute and replace.
There is another layer we simply have not talked about. Such tasks have sort of stayed under the bonnet. They have not shown up in any statistics, because nobody was fired. And we do not see anyone having stopped being hired. It is simply that in some segment, for certain tasks, a person left for one reason or another, left the company. Snap! And those particular actions were distributed among different employees. Or certain actions were handed, for example, to a model. So there is no answer — no single answer — to this question.
We have already listed three components. I think the list can be continued. So my answer is this: to the question 'will AI replace us?' — no.
In some tasks — yes. The figures we looked at, using the US market as an example, do not show any general wave of mass layoffs because of AI. They show and highlight a more complex process. About caution in hiring, about the cutting of some roles and the non-replacement of them, about the cutting of some roles and their replacement with new ones, about the change of tasks within a profession, about the caution, perhaps, of people somewhere in this uncertain reality. If it would be interesting to take apart similar figures for the European market, or the Russian market, or Asia, please write in the comments — we will definitely do it.
But in essence, today we talked about the market starting to look for specialists who understand their work so well that they can decide, overall, how that work should be, how it can be changed, which process can exist and which cannot, which part of it to hand to AI and where to keep the human. An amazing time — we live in interesting times.