Mateusz Madejski, Business Insider Polska: Everyone has been talking about agentic AI lately. Does this really tip the table over?
Piotr Kawecki, exeAI: The automation process is actually happening – and it can be seen, for example, in Polish industrial companies. We have more and more digital twin solutions, i.e. simulating the production line using AI tools and conducting experiments and changes. This, of course, speeds up the entire process fantastically and reduces costs – because such experiments do not have to be conducted in a real environment. On the one hand, we have more and more interesting solutions, on the other hand, we have more and more maturity to implement them sensibly, and on the third hand, there are KPO funds that can be used for this purpose. This creates a very hot crucible. I believe that thanks to this, many interesting AI projects can be born in our country.
“The agent will do it in a few minutes”
You have something like this on your website – “intelligent real-time CFO”.
Jarosław Sokolnicki, exeAI: When it comes to controlling and finance, there is a lot of potential for change here. The way it works is that the agent selects the appropriate algorithm and makes the necessary calculations. The agent understands the intentions we have in our organization – and builds everything needed to solve the problem. For example, an analysis can be created in just a few seconds. How would this happen without an AI agent? The financial director would go to the analytical department and ask, for example, about the margin if no promotion is applied if March is rainy. Analysts would scratch their heads and tell the CEO to come back in 2-3 days. The AI agent will do it in minutes.
Will a virtual CFO replace the real one?
PK: In the case of AI, one of the biggest challenges is the issue of control over what the AI does. Even in EU legislation there is the concept of “Human in the loop”, i.e. a human being in the decision-making loop. So at one of the stages there should be a human who will supervise the results generated by AI and make the final approval. Until technology reaches the point where we trust it like a human being, this human is crucial to ensuring – or confirming – the outcome of AI's work. So we do not want to replace the CFO, but rather limit the time he spends on analyzes to a minimum. To leave him with the very essence of his job – making decisions. Therefore, AI does not replace it but relieves it.
JS: This is primarily intended to be support for the CFO – to relieve the burden, not to replace it. Although in smaller companies that do not employ a dedicated person for this position, agents may actually take over this role entirely.
Is the business world moving towards agentization?
JS: I absolutely agree with that. We no longer admire chatbots, now we admire agents. This is truly a game changer.
PK: I also believe that AI agents can give many smaller companies that do not have this whole legacy, all these lawyers, accountants, CFOs who decide where the company's money goes. All these smaller companies will therefore be much faster and more willing to enter AI technologies and “employ” agents. Because the pursuit of customers, revenue and optimization is greater there than in large organizations. The question, of course, is how quickly they will react and how deep the changes large corporations will be ready for. I don't know the answer yet, but I am watching this phenomenon with extreme curiosity. There is certainly a good chance that AI agents will democratize the market a lot.
JS: It's worth adding that there is a big difference between using AI and real implementation. Most people on JDG use ChatGPT for writing contracts or calculations – and that's great. But to manage processes in the company, large implementations and dedicated engineering solutions are needed. Kowalski from ChatGPT will not do this.
“No turning back”
Are agents likely to encounter any obstacles?
JS: Technically, the possibilities are already there. This is more of a cost question. Most models are settled on the so-called tokens. And there is always the question of whether it makes sense to spend so many tokens to achieve the desired effect. Is it possible for AI solutions to become cheaper? Probably yes, but they are also very energy-intensive, so the bill will not be low. There is certainly a willingness on the part of management to experiment and pursue AI agents. And that's where we are – we help turn this desire into specific, measurable implementations.
PK: Everyone probably asks whether AI can replace it. Or whether he can replace his colleagues from work. However, I am reminded of the history of our civilization. Change has always generated fear – whether it was when the steam engine was created or when the automobile was created. There will always be those who don't want new technology, but change is happening anyway. I think it will be similar now. We have technology that is already ready, companies that are looking at the cost side – and in between all this there are us, real people. On the one hand, we want development, on the other hand, we are also aware that we are sawing off the branch on which we are all sitting. And you will simply have to learn to live with this technology. Because there is no turning back from AI…
What is the exeAI project?
PK: We combine two worlds – business and technology. Or maybe differently – we combine the latest available technologies with a slightly “old-school” approach to business – where the most important thing is to understand customer needs. Also those who are scratching their heads as they see what is currently happening – and what changes are taking place in the world around us, shaped by AI.
JS: We want to make our customers' dreams come true. Their expectations regarding AI are, of course, enormous, but we also want to make them more realistic. We have a lot of attempts in the market to use AI for optimization. However, the question is about skillful implementation – that is, how to do it. We have prepared a 4D methodology that guides us and the client by hand through these changes. We monitor all these processes and see them through to the end. We also have, for example, ROI Studio, which allows the client to calculate the expected return on investment. But sometimes these processes are only in the minds of customers. And this is where we want to help.
PK: We don't get excited about new LLM models, but we look at the real needs of customers. That's what I think sets us apart. AI technology is a great thing, but without selecting the right tools for the right needs, it is just a toy.
See also: I learned about the train production process. One fact really surprised me
