Actually, the purpose of our meeting with Serghei was to talk about his recently published book “Why Digital Advertising Doesn’t Work.” A fairly predictable interview in support of the book’s sales and promotion.

But we hadn’t seen each other for a long time, and the conversation began not with the book, but with a discussion of the hot topic of the new leap in the development of artificial intelligence technologies.

Here I will add from myself that throughout the whole of 2026, although it has only just begun, I can’t get rid of a visual comparison of what the exponential growth of artificial intelligence looks like. When water is poured into a glass, doubling its amount each time. While the water was just at the bottom, this procedure seemed very slow. Then, after another tenth or twentieth repetition, it finally reaches the middle of the glass. To the eternal riddle: “is the glass half empty or half full.”

And then just one movement and the glass is full to the brim and already overflowing.

I deliberately stretched out the description like this. To convey the feeling that throughout 2025 AI somehow no longer surprised anyone with anything special, and suddenly everything started pouring out, in completely different fields…

So it was perfectly natural to begin the conversation with Serghei with what feels the most acute…

Serghei, hello. Have you already seen OpenClaw, which was first Clawdbot, then Anthropic insisted that they change the name because it was too close to Claude, as a result within a month they were first Moltbot, and now they have become OpenClaw. If OpenClaw develops at the pace it showed at the beginning, then soon we will be able not to meet for business matters, but simply initiate a meeting between our AI assistants. They will discuss everything among themselves, for example the question of promoting your book and its best promotion, and you and I will only confirm their decisions. And we will chat over coffee about something not related to work.

In my opinion, you are still getting a bit ahead of yourself. But I agree with you, the AI world has accelerated again. The same Anthropic that you mentioned recently released updates for Claude tailored for lawyers and financiers. This is a tectonic shift. Law and finance are extremely regulated zones where everything works according to formulas and codes of law. That is why AI is implemented there more easily than in “soft skills”. After the presentation of Claude’s legal features, the shares of five of the largest law firms fell by 15%.

Recently I contacted lawyers, I needed to draft a not very complicated contract. The work was estimated at more than a thousand euros. Of course, on the one hand I understand the lawyer as a specialist who invested years in education and practice. But on the other hand, AI with a subscription of 20 dollars per month has access to the entire legal database and, if it does not completely replace the lawyer, it allows a good lawyer to spend just a few minutes drafting a contract. When services are paid by time, it clearly is no longer worth a thousand euros.

Your forecast for 2026?

No forecasts. I can just show you, so that you smile, the website http://rentahuman.ai where AI hires people, meaning not it for us, but a human becomes an “exoskeleton” for AI, performing in the physical world those tasks that are still unavailable to code.

Here, it seems to me, you are already getting ahead of yourself. More precisely, not you, but the authors of the project idea.

Well, if we put jokes aside, we as a company have practically stopped following the growth of workload with symmetrical hiring of employees. There is no point in increasing staff when AI multiplies the productivity of one person. Now a junior-level specialist can work with the efficiency of a middle specialist, having access to support from AI. I am not even talking about seniors themselves.

Are clients already asking you for AI-based solutions?

Or we propose them ourselves. A simple example from recent practice. We had a client in the agricultural sector, a large company selling fertilizers. The specifics there are such: sales managers are experienced specialists, people in the 40–50+ category who have been in the fields for decades. For them any modern CRM is “Chinese grammar”. It is difficult and they have no time to sit at a computer entering the results of meetings and calls into the small windows of the program. Because of this the company was losing data, leads were forgotten, and analytics was “full of holes”.

A familiar situation. The reason why I still don’t have a CRM and instead have a bunch of Excel tables…

So we solved the problem radically we removed the need to type. We created a tool based on AI where an employee simply needs to make a phone call with a client. AI instantly transcribes the speech, extracts the essence from it (who called, what was agreed, what is the deal amount) and distributes this data itself into the corresponding fields in the CRM. Moreover, the system itself assigns the next task in the calendar. For a manager 40+ it became a salvation he simply “talked to the phone”, and all the bureaucracy was done for him.

A good example. Although from our previous interview I remember more your concept of work when you make a “No-Risk Offer”. That is, if your concept of promoting specific services or products does not give a return on investment to the client, you return the money. Though this concept is sometimes difficult to confirm, it is not always easy to measure the result of your work.

Let me give another example right away. We made an offer to promote services for a large medical clinic. They liked the proposal, but the same question hung in the air how will you prove the effectiveness of your proposal.

And what did you do? Especially if the service is not new, how can you prove that new sales are your merit?

We implemented a dynamic call tracking system. On the company’s website and in advertising materials different phone numbers were used for Google and Facebook traffic. The system automatically replaced the numbers on the site depending on the source of the visit. IP-telephony data was matched with the patient database. By the phone number, as by an ID, actually completed deals and payment amounts were tracked. This made it possible to prove that with expenses of 500 dollars per month on Google advertising, the clinic received revenue of 2000–3000 dollars per month.

I agree, this is a provable option.

In the most extreme case you can use the “Delta” method, when there is a fixation of the average product sales for a period without advertising (baseline) and then the measurement of sales growth after the launch of the campaign. This is how we worked with a new product from Lapmol. The effect was measured by deviation from the average sales value.

That is why most agencies sell reach, likes and beautiful reports. But clicks are not always equal to profit. At Five Stars we turned the model upside down: if the client’s financial goals (KPI) are not achieved, we return the money. We take responsibility for the final result.

Well, with more than 100 active clients, Five Stars has enough cases from you for more than one hour. Let’s still talk about the book, the conversation should be dedicated to it. I actually have one question. Does the value of a book that you prepared for at least a year remain when such crazy changes are happening around? AI versions become outdated in months, why then is your book still relevant? To the question that became the title of the book “Why digital advertising doesn’t work” new answers keep appearing.

You hit the point with the question itself. The book will not become outdated quickly because it is not about answers, but about questions. Consider it 35 chapters about questions that any entrepreneur must ask themselves so as not to simply “drain” the advertising budget.

Artificial intelligence has become a real exoskeleton and accelerator of our efforts. But what is the point if you determine the wrong direction and apply double or triple the effort to move in the wrong direction? You will simply go even further the wrong way. If you need purchases from a premium audience, then why are you chasing views of advertising material? Maybe you need not one hundred thousand impressions, but one hundred, but with the right conversion?

One could say that the essence of the book is that it helps to start understanding the mathematics of business. You always need to find that very point where you understand how much profit each invested dollar brings. So the book helps to identify systemic mistakes that “eat up” money even before the client presses the “buy” button. First we build the logic of actions, and only then we put on it the digital “exoskeleton” of neural networks for scaling, or plan this or that digital advertising.

Position accepted. Now it is clear what it can give, even if we start communicating at the level of assistants in artificial intelligence. Whatever answer technology gives, the question is asked by a person. So if the book is about questions, then my last question where did you put it up for sale?

In Librarius, Carturesti and Bookstore.

The last phrase is remembered best, so let it remain: “Why Digital Advertising Doesn’t Work” in Librarius, Carturesti and Bookstore.