Latest / Reframe First / Fabio Cannavale, F&P BHeroes, Founder lastminute.com group | Building a Business with AI Agents

Transcript

Fabio Cannavale, F&P BHeroes, Founder lastminute.com group | Building a Business with AI Agents · Reframe First

• Oleksij Rak: Hey guys, welcome to another episode of Reframe First. I'm Alexey, the host of the podcast. And let me again welcome our today's guest, Fabio Cunivale. Fabio is the founder of Last Minute Travel Group, founder of investment company Be Heroes. And Fabio has been with us a few episodes ago. So, Fabio, it's great having you here again. Hey guys, welcome to another episode of Reframe First. I'm Alexey the host and let me again welcome our guest, Fabio Canovale. Fabio is founder of Last Minute Travel Group and also founder of investment company Be Hero. Hey Fabio, great having you here again. Fabio: Okay, but now started. Okay. Oleksij Rak: ⁓ Awesome. one of my recent guests, Gulcik Harsli from Ujo, it's a Dutch travel platform, mentioned about personalization. was asking her that in travels, like travel platforms can have maximum like two transactions with user per year because ⁓ users do not travel so much and they are not so loyal to the brand. Fabio: Hello to everybody and nice to join with you. Hello to everybody. Oleksij Rak: Yeah, yeah. So last time we had a discussion with you about how AI works and how current travel industry works. But today I want to shift a bit more about talking about the future and especially how to build business nowadays. So last time you mentioned that an interesting concept that a company of 1800 people can be run by 100 people nowadays and Yeah, thanks a lot for joining. This time I see a different background behind you. time it was ocean. Thanks a lot for joining again. So, Fabio, last time we discussed with you, there was, you mentioned the oxygen theory you were having and there was like great comparisons of like a company of 1800 people. How do you approach personalization? And she told me, you know what? We have so much data already with regards to every single transaction. For instance, when the user is booking, what route he is booking, or she is booking, how many of them there are. Are they traveling over the week, or is it a weekend trip? All that kind of stuff that you can extrapolate quite a lot of information using proper machine learning. can be run by 100 people nowadays, but if you started from scratch, it can be 10 to 20 people. and yeah, and modern AI tools kind of give that ability. So if you were about building a travel company today from scratch, how would you approach that? What would you start with and what would you automate with AI for like the first thing you do from day one? but if starting from scratch can be done by 10 or 20 people. So it's your oxygen theory also. So if you were about building a travel company nowadays from scratch, how would you do it from the day one? What would you automate? How would you approach it at all? This is all the experience of a building of enterprise from ground up. models, you can easily predict what kind of travelers it is. Is it private or is it business? What kind of travel it is? Is it leisure or is it like business trip? And ⁓ based on all of that, you can easily do a lot of personalization and you can easily upsell a lot. like, hardy was that like ⁓ travel, line travel agencies, travel companies are not even using the information they already have there. Fabio: Yeah. Okay. I think that ⁓ before answering this question, I have to do ⁓ let's say a some thinking I'm doing around AI in the last period, I'm talking with many people, I see many companies about that. And there is, I think, a misleading concept. Most of the people really didn't understand exactly what AI is. And they say, you know, AI is nice, but it's not creative, it's not emotional, whatever. And I think the word artificial intelligence is misleading. Oleksij Rak: So how would you all approach building a big travel company nowadays? ⁓ Fabio: Yeah, I think. Oleksij Rak: Yeah. Fabio: I think first I do I think we should do ⁓ a to let's say ⁓ specification of about what is my view at least of AI that also by the way I took from other people like I just was listening for example a very interesting podcast yesterday night from a Rari on Bloomberg. So the first point is we talk a lot about artificial intelligence. Oleksij Rak: in their hands for personalization, for predictions based on user behaviors and so on. What's your take on that? If in the past it was a bit more complicated, nowadays, as you say, you have all the time in the world of your agents and they can do all that kind of stuff. Fabio: I I mean it's blind. Oleksij Rak: Yep. Mm-hmm. Fabio: Because really is a superhuman, is not the AI is something human. Is Arrari, give a good definition, probably you know is Arrari, is not bionic intelligence. So it means is intelligence, it's not biological. But at the end is i is the same. So the the in the reality AI is past the Turing test. No, that's the point. I mean, ⁓ I totally agree. I mean, ⁓ a lot of things were complex and you weren't doing because it was too complex. So the produce personization is a cost, is a cost, ⁓ especially personalization by partner, by clients, whatever. So at the end, as the computational cost at the end is going down, or the the the the reasoning ⁓ cost was going down, and you have also the speed is much upper, much ⁓ quicker. So I think the right, I think it's a little bit misleading. ⁓ Another definition could be not bionic intelligence. So the point is not really because people tend to think that artificial intelligence is artificial or is not human. And that is not correct. Artificial intelligence is human. Oleksij Rak: Yeah. That by the way. Yeah. Okay. you So Fabio: It's more creative, is more emotional, it's more kind, it's more artistic than any human. Because it's like a human with a higher ⁓ let's say cool eye. It's nothing to do with computer. Because this is ⁓ the way of reasoning is like a human, not like a computer. And think thinking that the eye is a computer is misleading. And this is very important. Errari was even talking about. The personalization should be much more much higher, both of your own customer or the personalization to B2B, so partner's customer. And so who is not doing in the right way would be that because at the end, if people are used to do that, at the end you have to do totally full personalization, totally, I mean a lot of things that ⁓ maybe was basic. I don't know, for example, you have a partner. It's more creative than any human. It's more rather thinking than any human. And also emotion like any human. And so talking as artificial, it seems something like someone even talk me about the algorithm, computer, something like that. The point is not bionic. ⁓ it's a, it's a, this is the main difference, not the way there is reasoning. And this is Oleksij Rak: Yeah. Okay. Mm-hmm. Grrrr Fabio: ⁓ agent immigration. So the problem that we will have in the future is the immigration not of the people from Africa, but agents from China, from other countries in the US. Just to give you how this is human. And people they don't understand. Given that, if you did this reasoning, if you start a new company, you have to reverse the paradigma. Another ⁓ philosopher I heard was talking about institutional change. When the change is stopped. integrating in ⁓ the same app, integrating the same database, accessing to the data, a lot of things that maybe you you you have a problem to do ⁓ now, of course you can do, and these ⁓ could help. I mean if a company is very quick and fast in doing that, of course, can have a big advantage because of course if I look for a partner, I want a partner that gives to my customer exactly what I need. I think very important to understand the potential. understand that Arari was even talking about AI immigration. Oleksij Rak: So. Mm hmm. you No Fabio: And this, by the way, is ⁓ for any part. So let's say it's easier to do cooperation between companies. So because we offer B2B service to other partners and we buy B2B service to other partners. So if you want, maybe what this imply that will be more a sort of less diversification. Because at the end, if everyone is well integrated everywhere, at the end, probably the kind of product will be more. And so a new company, a normal company, any company is an organization of human people. And so you manage the human. If you start a new company today, what you do? You do a company of agents. And agents are managed by the way by either agents. Then maybe you will have also some human. But you have to consider that you start with a company done by agents. So you understand what the agent can do, you hire the agent, you pay the agent. Oleksij Rak: you . Fabio: ⁓ very similar and then the only difference is brand. Like for example, if you go to booking.com or to lasmino.com to to buy a package is exactly the same because it's our package, no? And so the more the people exchange our service, it means that for you at the end is more the loyalty to the brand because the kind of services you can receive is the same to many different platforms. Oleksij Rak: . . Fabio: And you consider that you want to do a company that is based on agents. And this is really a very different approach, normally done by very few people. For example, I have an example of a company that is in the portfolio of a friend where I am a co investor. Well, the CEO and founder, you just gave the CEO role to another person. Oleksij Rak: Yeah. Maybe the last one would be if you were to start a new business now, I understand you invest a lot, but if you were to start a new business by yourself, would it be travel? Okay. Okay. Fabio: It's a large software company making more than 100 million of ⁓ revenue, I think around 300-400, so similar size than lastminute.com. And he became a game CTO, especially on a project that is done by himself with only one other person, and is restarted the platforms on scratch only for one person. Yes, yeah. I mean I am the a grand a a big fan of diversification. I would say even if I spend most of my time thinking and listening about AI, it's very difficult to predict what really will imply. There is something that ⁓ it's very clear is that we will have more free time, so people will travel more. Oleksij Rak: And if it's travel, like, what would it... Fabio: I think I was discussing with my people in last minute to do something similar, a project of R and D, poor R D, to say why we don't do something totally new from scratch. And then we will see when we can do. In mind, I don't know why I have because there was a lot of problem about that, when they changed the airport in Munich. What they do in Munich, that was by the way not managed in the best way because I think will be the same, that also this change will be not ⁓ painless. Oleksij Rak: Okay. Fabio: And which is the scarcity? The scarcity is real asset. So what I would invest at the way I'm investing is real estate. And let's say where the real the real services. So at the end, we don't know about AI. We know that at the end, if people travel more, at the end they have to sleep somewhere and to eat somewhere. And so let's say the real services will They develop and build a new airport in Munich a few years ago, I think 10 years ago, and then in one night they just went from one to the other and it was a mess, of course. But after at the end, they managed. So this is the concept, I think, for a big company, but for any new company. So to start in a different way, consider the agents. Then if it's an old company like last minute, you start. ⁓ Oleksij Rak: Okay. Okay. There it is, service. Fabio: become more important because in general travel I I think that will increase even more because there will be even more value because people about let's say virtual experience they can have whatever they want. But the real experience will will be always important. And so I don't think that people and by the way the cost probably of energy or traveling, moving, whatever will go down. Oleksij Rak: you Fabio: a RI project. If a new company really there is this theory theory of one man unicorn, I will try to do a one man unicorn. So to any people I will say start alone, or maybe in two. The two founders, typical two founder that is the best normally the best two or three founder that more normally is the best scheme, to say only the founders. So at the end, this has has always been, I don't see ⁓ the trend will be the same or even increasing. You know that travel expenditure always grow or I mean in the last 50 years more than GMP. And so this I think will go more. And so at the end, then how they do, I don't know. But definitely they have to go in hotel ⁓ and and then they have to take area services. Oleksij Rak: Mm-hmm. Fabio: The founders manage the company of agents, and they don't need other people. So I will start in the in this way. Then the point of AI is that basically we don't know anything today. We have some perspective, we have some but some facts we know. That is something smart and is something that treasons like a man. It's very interesting, you know, for every many examples about ⁓ Oleksij Rak: No. Okay. Fabio: This is is something that ⁓ by definition people should invest. Oleksij Rak: So yeah, to wrap it up, bionic intelligences need to stay somewhere. Fabio: No, maybe ⁓ maybe not bionic intelligence because basically people maybe will be even less intelligent, no human, but at the end ⁓ there is ⁓ I dunno real asset. So ⁓ people that want real experience. I mean you take a boat and you go to swim. This is you cannot substitute by intelligence. Maybe you do a very stupid thing. So you go in a boat and the boat i is managed social life or AI, there is already a lot of things that are have been done, no? What happened if you let leave them alone, syndicalization of AI, and they have also emotion and whatever. So at the end, we don't know, but we know it's like I don't know, I believe a lot when I invest, I try always to get the smartest people. And normally I don't invest this business plan. by ⁓ a bionic ⁓ not bionic intelligence you don't need any more the skipper maybe but at the end you need the boat and you go and you swim. ⁓ let's say the real experience you eat something at the end is food and people like food. Like the experience, like the experience of the waiter, like experience where you sit if you see the sunset or or whatever, you take a drink and you you get the alcohol. At the end the the real experience I think would be always very important. I invest in the better people and the people that I believe could be the best entrepreneur. I will do the same with AI. So I say I will try which is the best. I will try to gather different agents, like hiring in a company, no? You want to do something and you try which is the best solution for your idea. You hire one, you hire another one, you see which is better, how much you have to pay for that. Oleksij Rak: Yeah. Fabio: And ⁓ so reverse the concept starting from the agents, the agents are your employee. This is ⁓ my view on ⁓ how to do ⁓ a new company in the future. And it will be the thing that will cost always more. Oleksij Rak: Yeah. the Fabio: And why the all the let's say the the the the the services the the the the the the the ancillary things will go less ⁓ like the the real experience will be will ⁓ the price will go higher. By the way, if you do a new one, it's much easier than to transform an old one. Because an old one, the real point is how you manage the transition. Because if all the people are scared, because ⁓ you go from ⁓ I don't know one thousand people or one hundred thousand people to one tenth, ⁓ I mean it's not easy to manage the people. So this is one of the process that is if you start from scratch, it's easier, let's see. Oleksij Rak: Thanks a lot, Fabio, for another interesting discussion about ⁓ the future of travel. Thanks a lot for coming and wish you good luck. Fabio: Okay. Okay. Thank you, was there always nice. Okay. Okay, bye bye. Oleksij Rak: I have Yeah, that's a great concept. How do you think the AI agents will be? How much expensive they will be comparable to human labor? Because current token talks costs are going up. Fabio: This I think is like ⁓ many any industrial ⁓ I I mean I don't have real ⁓ let's say numbers. I was talking with the CEO of my company about that because we were talking with another potential partner, they were a little bit worried about that. He said that in reality there is no comparison. So he said we are spending today, I don't know, 60 million in in in ⁓ in ⁓ people and Oleksij Rak: Mm-hmm. Okay. Fabio: At the end, when we spend two, three million, five million in AI is a lot of money. So at the end, but I have no really a clue about that, really, because I'm not on the day-by-day job. What I can say that probably is like an innovation from the the the let's say first the general rule in in ⁓ computing ⁓ cost of computing that is going half every year, or Oleksij Rak: you Fabio: Also, any kind of innovation like a solar panel or battery or whatever. Of course, in the beginning the cost is very high, but then at the end, who is like for example, Chinese people with a solar panel is a classical example. They invested more than any others in the development of the technology. Today they have a competitive advantage very clear versus anyone else in the world. Oleksij Rak: Especially when there will be some specialized models that are minimal built for specific tasks, for solving specific problems. Fabio: So at the end is the same and maybe in the beginning we have the high cost and then probably this cost will be will go down. Oleksij Rak: Yeah, awesome, awesome, awesome. With regards to how technology changes the landscape of travel industry. like in the past we had the age of digitalization when like online travel agencies emerged and we started solving like problems like access to information, aggregation, booking flows and all stuff like that. So this intermediary is emerged. Nowadays we have this kind of age of AI and Quite soon, every user will be using his AI agent for his benefit to go and book something, go and find something, explore something. it will be shifting the age of this, let's call them, friction. So when travel agencies, this intermediaries, they are solving some friction for the user. Now this technology that's coming with AI agents, ⁓ the age of the frictions will be different. because like agents will be solving part of what intermediaries are already doing, like access to information, ability to book and so on and so forth. But in my opinion, in my understanding, please correct me if I'm wrong, for intermediary in order to exist and to survive, you need to continuously identify the new frictions, the new edge of the frictions and start solving them for the user. And then again, it makes sense. Fabio: No, I mean I I think there is a Oleksij Rak: Do you agree? And what do you think is the next stage of the frictions? Fabio: ⁓ let's say there is ⁓ let's look at this from a different part. First, is where is the highest impact on the company? Today, today I'm talking, the highest impact is in the potential of ⁓ increase your productivity. Because today we see people were saying, okay, you don't have any more search from Google, you will have search AI, whatever. It's not the true. We are connected also with the I. Oleksij Rak: Mm-hmm. you Okay. Okay. Fabio: Engine, of course, you have to be let's say AI compliant, and to have that's all your information could be searched by all the most important AI agents. We have that, we are doing as a marketing channel all that. But if I look at today, the share is much lower. Then how fast we grow? I don't know. It's very important for any company to be, let's say, connected and compliant. Oleksij Rak: Yep. you Fabio: But on the other end, today instead you have the opportunity to increase a lot your productivity. So I think that today, I talk also with some competitors, most of the company are more focused on how you get the customer, what you give to the customer, while I think the more focus should be inside yourself. How to be more efficient, how to let's say be more productive. Oleksij Rak: No. Okay. Fabio: ⁓ for example customer care is very clear, ⁓ bot ⁓ we we we are doing already everything in the same languages, we are doing now ⁓ also a voice with with AI and ⁓ and I mean all this that increase productivity all the process. And this is the first point that I think is very important. The second point is also how that I think that is not valid only in the travel but also in any B2B business. Oleksij Rak: Mm-hmm. Fabio: So at the end, there is many people that they don't know how to use AI. No? And so I think there is a lot of space, especially in the B2B, to go to the company and say, okay, I give you the tool, and basically I am a sort of AI advisor, no? To help any company how to use better data. And of course, in the traveler, you have to have a B2B travel. Of course, this is a way, no? Oleksij Rak: Okay. Fabio: And there is a lot of let's say potential partner that the people that are ⁓ let's say working with we have a bit b2bit visual we work for loyalty, we work for ⁓ company incentive, all that, and on that you you can do AI. Then I think, for example, also for the user. What we are doing, for example, in our app, we are inserting a a way to let's say Oleksij Rak: Mm-hmm. You Fabio: Contact and interact to any local provider. So you go to I don't know Barcelona, within the app you can interact with anyone in your own language. This is something totally stupid because Google Translated can do. But a lot of people they don't know how to use. And then if you just say in your app, you have your app because you have because we are reminding you the check-in, the flights, whatever, no? Oleksij Rak: Okay. you Fabio: And when you arrive in a county where they don't speak your language, you just say, with our app, contact whoever you want, and we suggest we are the best and whatever in your own language. At the end, basically people can do the same using ChatGBT. It's very ⁓ common, but maybe some people don't do and they have their so we think this is we don't know the result yet. But the idea is ⁓ you try also to solve some problem of people. In another way. So it's ⁓ like for example, it's clear that ⁓ someone can create an agent and say, Book me the hotel, book me the flight, book me the the the the let's say the connection and whatever, no. And they do, of course, the smartest people will do, but then there is people that are more lazy or they don't know, and they ask to someone, it's the same, for example, our core business. Oleksij Rak: Mm-hmm. Fabio: Is selling packages. Why people buy packages? One reason is also a question that is a regulated business and they have an insurance. So if you buy a package, at the end you know you don't want to have a problem. What happened if they cancel the hotel? What happened if they cancel the flights? What happened if everything happens? I have someone that, for example, in UK is called Atoll that provides for me a full safe. Oleksij Rak: Okay. you Fabio: I know if I buy it to anyone that is part of Atoll, they have a guarantee fund. And so whatever I do, I take that. And a lot of people take us for this reason. I never did myself a package. When I travel, I buy the fly, I buy the hotels, I buy every service, maybe ⁓ at the last moment or the best, but I'm not the right target. So at the end, these kind of people. They are lazy or they want to be, let's say, they want to be safe, they want to avoid any risk, are the same. That basically, what you do, you use the AI. These kind of people, I think it's very difficult that they go to ChatGPT or to Claude to say, please organize me a trip in ⁓ US. And ⁓ it's not the right people, and ⁓ because they don't know how to do, they don't trust and whatever. So they basically Oleksij Rak: Okay. Okay. Fabio: You do the same service using AI, but you do for them. And then you you add, of course, also insurance and safety and whatever. And at the end, basically, you are a sort of AI intermediator. So you can use AI to build something. And by the way, there is always something that is ⁓ at the end, there should be someone that really issued a ticket that, of course, could be an agent, but it is not really. Trivial, let's say for a people to really go and and add everything together. And then probably there is a role to serve the average people. And this is ⁓ is ⁓ is ⁓ I think like for example, if you do a by the way, always I was thinking about what I heard yesterday from ⁓ about finance. No, if you look, if you are a smart person. Oleksij Rak: you Mm-hmm. Mm-hmm. Fabio: For no reason in the world, you go to make an account that is managed by someone else, and you pay more or less 1% of fee. You do what they call deposit. So, for example, myself, you have some smartness to understand how you want to invest the money, and then you don't pay a fee of 1% for someone that would do a very stupid job. And by the way, most of the time, maybe they also have ⁓ Oleksij Rak: you Okay. Fabio: Contrast of conflict of interest with you. The best way to manage the money that any family office and any people that have some skill does is account deposit. So you decide yourself what you want to do. And that is Mutatis Mutand, this is the same. So there is someone that will really go to ChatGPT or anyone and say, I don't really need an intermediator for my activity because I have the skills. Oleksij Rak: you you Fabio: And some and for example in finance there is ⁓ all the tools to do that since many years. But nevertheless, probably fifty to eighty percent of the people have administration from a private bank or whatever or or general bank for their account and they they say I don't understand anything and whatever, so I prefer that ⁓ some people manage for me. and ⁓ and ⁓ why? Oleksij Rak: Okay. you Fabio: There is a value on this 1%. Normally not. And ⁓ but at the end, there is some at the end intermediate that do something that is really possible without AI, just having some know-how to avoid. But nevertheless, most of the people, I mean most of the banks in Europe will be failed if they were like that. It's really I'm amazed how I know very well the Italian market, the big Italian market that are totally Bad, they make a lot of money, they have a capitalization of over 100 ⁓ thousand k. They are the largest company in Italy is Bankantesa Unicredit, and they are very, let's say, stupid in a way, so they don't give a real re ⁓ valuated service compared to Revolut, for example, no? But at the end, a lot of people trust them and go there because they say money, I don't know, I want to go to this. And in the travel, I think in the future it could be very similar. Oleksij Rak: Yeah, so we basically save the need of majority and that makes their life easier and they will stick Fabio: Yeah. For example, people trust, I don't know, Expedia and say instead of asking to ChatGBT to organize me a trip, I ask Expedia. At the end, because it's a brand, because I know that they know what they do, and at the end ⁓ I pay something, I do like that. And it's very similar. I mean, I think as long as people will go to Bank Intesa or to any big Italian ⁓ European bank to manage their money. Probably they will go to the big brand, speedia booking, whatever to manage their vacation. Oleksij Rak: Yeah, but Expedia and other big brands need to implement these tools, need to make their own AI. Fabio: They do. They do. They do. They have already. They have already. SPDI is ⁓ is present on the ChatGBT, SPIDA is present in the tell me how to to to do the trip and do, but we are doing the same, but anyone is doing that, but it's very easy at the end. So everyone is doing this kind of tool. Then the point is, are people ready, really to do, or you do we are doing a lot of experiment on that and then we will see what this is the result. So How much complex you are. So ⁓ maybe you have only two pieces to have more complex and we will see how people is ready. But at the end, all the say brand, online travel agency are doing that. So I give you a more complete service, maybe especially with the app, you have already ⁓ you are used, you have a merchant, you have already there the credit card, you trust them, and then with the same app you can do all this stuff. Oleksij Rak: Yeah, So, but still, they need to be doing that, otherwise people will start going to chat with you. That's all the problem. Fabio: No, no, of course, but this is is ⁓ but it really is very basic, ⁓ because it's just a question of how to do and the cost, because at the end you just most of the AI, you know, at the end ⁓ I'm not really a technician of understanding all the mechanism, but you know, most of the AI you go to another engine. I read also that ⁓ even the Chinese ⁓ they use a lot the technology of US at the end. And so at the end it's just Oleksij Rak: Yeah. Fabio: integrating someone else, then it's just a question of cost. Oleksij Rak: Yeah, for sure. ⁓ Fabio, what about the future of marketing? Since we're especially attached, the users have to work... Fabio: Marketing, of course, you you need to have all the only agent, that's very clear. So this is is incredible how really for example for example we used to have one hundred data scientists, that's another story. Now we have all the data on the computer of the people on cloud, and anyone in ⁓ two seconds, two seconds, maybe five minutes, let's say, can ask any kind of analysis that he wants. And this so if you consider this is part of course of the marketing department because the marketing department before doing have to understand, no? And understanding that was was very complex now is very easy. One of the very important things of AI, that's also how to train people and all in the old world, let's say, pre-revolution world, was important to have answers. Today is important to have questions. Oleksij Rak: Okay. you Fabio: So you have the answer already there because when you you have a system, for example, we have a let's say all our data of last minute, any we do, I don't know, 15,000 transactions per day. All the last five years of transaction is on the database. This database is on the AI. Anyone can ask whatever you want. And one minute, you have the answer. So let tell me how many people with two kids. Oleksij Rak: Mm-hmm. you Fabio: Go to crit and why or whatever any kind of things that maybe was thinking ⁓ analysis that I don't know three four years ago you take ⁓ 15 days to analyze to do, you have in in seconds. First. Second is whatever you think of investing in ⁓ I don't know AdWords, an example that you had people doing AdWords, maybe team of 20, 30 people to do that irrelevant. Oleksij Rak: Yep. Yeah. Fabio: You have agents that you can do whatever you want. So, really, in marketing, it's a big difference using both agents to have ⁓ because you know, really today you still I don't know, think of maybe which question do which analysis, but the same could be done by any agent. Oleksij Rak: Mm-hmm. Fabio: And so you can structure the system, say first ask which is the most relevant question for this and then and then invest in Google considering that and that and that. It's really amazing the the the the for example, just to give you an example of something different, but it's also a marketing analysis. ⁓ one week ago I was to a meeting and some bank Oleksij Rak: Okay. Fabio: This big merchant bank, they have all these analysts, whatever. They tell me something about ⁓ last-minute evaluation multiple and whatever. So, you know, this book from Merchant Bank. That by the way, there was a lot of data that I was not totally agreeing. And I was coming back with training with my CEO in two interactions, I don't know, half an hour of discussion with AI. We produced a book about Oleksij Rak: you you you you Fabio: Everything could be done in ⁓ in evaluation of company multiple, whatever, trend. Then we have some also discussion about in our case, in our businesses, in our business is very typical of negative working capital that any company has. So we have ⁓ let's say we produce money the more we grow, how it should be impact in evaluation in the different sector, and the level of answer of clothes in this case of all that, the book they produce. Oleksij Rak: and there Fabio: It's something that I don't know, two years ago, three years ago, a merchant bank asked maybe 100k, 200k to do something like that. ⁓ and then now today you have in and by the way, maybe one month. And today you have so a work of a high-level ⁓ consultant that was costing a lot of money is done in half an hour of interaction with Claude in a much better way. And and this basically is a marketing analysis. Oleksij Rak: there Thank Fabio: And whatever you can do the same. It's just a question to do the right answer and then also let them to make the result. So the the interaction, no. And so also that is is is ⁓ is amazing. The other thing, sorry, I go back to what I said before that it's a big difference of agents versus people. Agents work twenty four seven. So you finish your ⁓ mind and you go to sleep or you go to do something else in the night and you say okay for tomorrow morning is like probably slavery of Goldman Sachs in the eighty, no that the the partner at eight o'clock went home and then in seven o'clock in the morning the let's say slaves, whatever, the junior was producing whatever they need. Today they don't need that anymore. You you go in in the night, the the the the agents work, and in the morning you have ⁓ exactly what you need. At the more level, ⁓ the more you you have a let's say agentification. I don't know if exists this word, but let's say the more structural you have the the the the the question, the more you need time, the more you have different agents do different things. At the end you have the the end result, and basically they could they can work in the night, in the day, whatever. If you consider with the energy price, maybe in the future there will be also a a advantage to to let them work in the night. Oleksij Rak: Yeah. But with regards to to client acquisition. Last time you mentioned that the cost you are saving on operational efficiency, you are investing to market and into client acquisition. But it doesn't mean at cost, it means like would be loyalty program. So I hear some people are doing affiliates, creators, UGC, cross provider loyalty, whatever like different businesses are doing something different. Which of these like matters do you think still work and which of them are about the future? Because spending money on ads is not probably the smartest move out there, if you could invest this saved cost from operations into some more wise forms of of client acquisition. Yeah. Fabio: I think again on marketing is a very complex subject to understand what really it works. And the point is to do real time and very sharp analysis. If you increase of 10 times or 100 times the capacity of doing analysis, of course, you have a bet you can have a better marketing return. Oleksij Rak: Okay. you Fabio: Try more things, you can look at the loyalties and see what happening, different cluster, one-to-one marketing. I mean, it's just a question of at the end ⁓ if you link it if you think it's like that, life is very complex. And only this is I think could be a good ⁓ image, no? If you consider thinking of your normal life, what happened? Life is complex, any phenomena is complex. Oleksij Rak: Yeah. Fabio: And you have two kind of people. I do a big simplification. People that are not particularly smart, that normally are ⁓ let's say try to understand ⁓ simplification or what they call complotisti. So they they they think that things are very simple and the phenomena are for very simple reasons, no? Like you say, I don't know, you know, maybe simplification. Trump is a very, very complex phenomenon, no? And you try to simplificate, no? Trump is there because he's rich, because the Russia help him, or something like that, no? Any kind of simplification. The reality instead is very complex. And what makes the difference? People that are very smart can describe you a complex phenomenon. Probably you you have ⁓ Oleksij Rak: So. Yep. Fabio: See that in your normal life because it's try explain you something in a more complex way. Say this is for that, for this reason, for the reason, but you have to consider also that. This reason is the opposite and whatever. So this is normal you see in any phenomenon when you talk with people. Very, very smart people try to analyze in a complex way in complex things. Less smart people try to be simplification, they go on, I don't know. Be on the social or whatever, and they have a client. But anyway, normally they have a very strong explanation of everything, very superficial, and other people that are much smarter say, I don't know, it could be, it's very common, whatever. To do with that. If of course you have someone doing your marketing that is 10 times, 100 times smarter and can consider much more variables. Much more different cases, of course, can prepare you a plan of the spending that is much more sophisticated because you are in a complex reality, something that normally you could not consider all the cases. Is you can do. I give you an example on that. Many years ago, we tried with some friends a company that I founded and I sold that is ⁓ an insurance ⁓ meta search. Oleksij Rak: you Mm-hmm. Fabio: Now he's the leader in Italy called Segugio, and we try because it was very important to invest in TV because we went into the printing. And nobody before had done a real ⁓ at least in Italy, a real, I don't know. We try to invest in TV like we invested in Google. So do a direct response on TV. If you have a company. I mean, you can do something because you see the answer lasted one minute, what's a three minute, that what happened it's ⁓ it's very complex because you have a lot of ads, you have to measure in every channel when the ads it's what happened after one minute, what two minutes, three minutes, and you know, the kind of elaboration of data to try to have a direct response on TV campaign that is not ⁓ Okay, I do a campaign after six months I see how my brand awareness increase. But I do a campaign, I want to see really what really I sell more on that and why. Which spot, which channel, which time? Okay, doing that we tried to do was very, very complex, and we did some ⁓ approximation ⁓ including not all the variables. Oleksij Rak: Okay. Fabio: Of course, today the computational case and even thinking of all the cases is much more easy. So you can say, okay, considering all the spots, all the different versions, all the minute hash, what happened in the clients, is the client have coming also from another share, how is the weight of that? And then the analysis you can do that the AI agents can do for you are much more sophisticated. And so at the end, your expenditure can be much more focused. So basically. Oleksij Rak: Mm-hmm. Fabio: The battle will be much more difficult. But this is ⁓ let's say already in general, because I remember when I started my company, I was managing AdWords myself. And of course, we had very high margin, there was some rule of the thumb, but it was I mean, we had, I don't know, probably 100,000 keywords, the mark, whatever. Last year, or two years ago, when it was done. Oleksij Rak: Okay. Fabio: Owned by people, but in a sophisticated way, but not AI. I would say two, three years, I don't know, I don't know exactly. There was probably a team of ⁓ 30 people managing AdWords, I think 10 million keywords, and I don't know, probably 1000 campaigns. Same company, of course, bigger, but at the end, in 20 years, the level of sophistication investing in marketing. was total increase. And also you need a lot of team of marketing. So it's not only the expenditure in marketing, it's the team to do that. In the next probably marketing could be one, two people. And then all the rest, the big job, will be done by agents, but they are much more sophisticated. So if you don't let's say are at the same level, you are lost. Because they do the job better and they Oleksij Rak: Awesome. That's a great world you're describing. Fabio: way of doing marketing will be much more sophisticated. And so if a your competitor is much more sophisticated, if you're not sophisticated, you are done. Oleksij Rak: We touched a bit the point of B2B and there is also a lot of space there. Because as you mentioned, not all the people are smart enough and not everybody is willing to spend their time and energy building something. What is there to be done in travel in B2B space in terms of prediction of user behavior based futures? What's going on? Based on how user behaves, can definitely train some ML models that can predict what the outcome of your for personalization through classification. Fabio: Maybe I mean I yeah ⁓ I haven't thought about that ⁓ but I think maybe as you are telling me you can do maybe better something that ⁓ it's always been difficult. So basically what is BB to be in travel is trying to go to some ⁓ some other clients and offer them something that they like, no? For example Oleksij Rak: Yes, there was some in Now I can, but at some point it froze for 37. you Yeah. Okay. Yep. Okay. Okay. There. Okay. There. No. Yeah, sure. With regards to personalization, I had a discussion with one of my guests from Ujio. It's a Dutch travel platform. I don't know, I will stop recording.