Webinar: Digital Brain, Human Brand

Digital Brain, Human Brand

We live in an era where marketing can no longer be understood without Artificial Intelligence. The "digital brain" promises efficiency, scale and precision, but can it keep the soul of brands alive? In this webinar, we explore how to balance the power of technology with human authenticity, and what role creativity, empathy and intuition still play in an increasingly automated world.

In this webinar, discover:

✔ The fusion between artificial intelligence and human intelligence in marketing;

✔ The role of empathy and creativity in an era dominated by data;

✔ How to preserve brand authenticity in an increasingly automated ecosystem.

(00:00) [Music] She has a degree in organisational communication from the Polytechnic Institute of Coimbra. She is a specialist in several executive programmes, including Customer Centricity and Business Analytics, and teaches at reference institutions.

(00:36) It is a great pleasure to have him as a lecturer in our master's programme in Marketing and Technology. He is also the author of the book From Zero to Hyperpersonalisation, correct, Tiago? It has been launched and he is a major specialist in this field. It is a pleasure to welcome him today to hear him speak about the future of artificial intelligence in marketing. Tiago will now make his short presentation and then we will have a debate at the end.

(01:04) Tiago, over to you. Fantastic. Thank you very much, Ana, and thank you to everyone who has joined us. It is a pleasure to be with you at the end of a long afternoon, for those working after hours, and in this case not exactly working, but joining us to share a little about artificial intelligence. For anyone on holiday who has joined us, I am even more honoured.

(01:29) Ana, it is truly an honour to be part of this group of lecturers in an executive master's programme that reinforces and brings capabilities to marketing professionals that are becoming increasingly relevant. At Liminal, we look for professionals with this type of skill set, and they are very hard to find.

(01:56) The times we live in are increasingly populated by technology. That is what we will talk about over the next 30 or 40 minutes, and then we will have a conversation and hear from you as well. Digital Brain, Human Brand is the theme of this session. About me, as Ana said, with one or two additional notes: at Liminal, we are a team of executives and professionals in marketing, communication, business management, IT and technology, implementing core technologies for customer relationship management.

(02:39) We work with platforms such as HubSpot, Salesforce, Zoho and others, in digital transformation projects centred on CRM, automation and business intelligence. We have been doing this for more than eight years with clients and brands that are very familiar to you. And as Ana mentioned, we recently launched the book From Zero to Hyperpersonalisation, where we connect marketing strategy, marketing and sales operations, and the way all of this can be brought together in technology platforms, including the use of artificial intelligence.

(03:13) The goal is to progress towards personalisation, or even hyperpersonalisation. I want to begin by telling you two stories. They explain why we increasingly have digital brains inside our companies and brands, and why it is therefore increasingly important to have human brands.

(03:38) First story: the story of RedBalloon. In Australia, there is an experience and adventure company, similar to the kinds of gift-experience brands we have in Portugal. They sell experiences that take place exclusively in Australian territory and nowhere else.

(04:01) They began to face a typical problem for companies in this type of market, or in markets with low barriers to entry. Suddenly, they had a huge number of alternatives and competitors. Since one of RedBalloon's largest cost items was advertising and paid campaigns, Facebook, Google Ads and so on, they turned to artificial intelligence to understand how they could optimise that budget. They chose a platform called Albert, an enterprise-level platform, not necessarily accessible to every SME, but very strong and focused on that specific task: optimising keywords and advertising inventory.

(04:42) Shortly after artificial intelligence had been put to work, a problem appeared and they had to stop it. Why? Because the AI started buying advertising space for people outside Australia, specifically in the United States. That did not seem to make much sense. They had no offer in any territory outside Australia.

(05:13) So what was happening? Artificial intelligence had discovered an unexplored and highly profitable segment. Not tourists, as we might think, but relatives, immigrants and Australian expatriates living in the United States, who had a very specific job to be done: finding gifts to offer family members, friends or loved ones in Australia, gifts that would naturally be consumed in Australia.

(05:54) This generated an extraordinary increase in the company's revenue. Explained like this, it is easy to say, wow, incredible, obvious even, but it had never occurred to RedBalloon's strategy or marketing team that this could be such an interesting and disruptive segment for the business.

(06:18) The second story is much closer to us. You may remember that last Christmas we saw the classic Coca-Cola Christmas ad, which was entirely generated by artificial intelligence and paid tribute to one of Coca-Cola's most popular and classic Christmas ads, the trucks, as you can see here.

(06:43) They did this entirely with generative AI. If they had used traditional cinematography techniques, the investment would reportedly have reached many millions of euros. According to what has been reported, Coca-Cola invested between 300,000 and 500,000 and a few weeks of work to reach that result.

(07:13) I am going to share four principles with you. The first principle I would like to reflect on is this: when we look at these cases, we may think that these results and technologies are only accessible to companies with many resources, the Coca-Colas and RedBalloons of the world. The first principle is exactly the opposite. This is not only for the rich and large.

(07:39) Probably for the first time in the history of technological evolution, the frontier, the most sophisticated technology available, is accessible to any company. In this case, we are talking about artificial intelligence. For example, if we wanted to replicate the RedBalloon case, there are several alternatives. One of them is called Zuma. For around 60 euros per month, there is a platform built and dedicated to small businesses precisely to do what we saw in the RedBalloon case: optimise online paid advertising placement, discover unexplored segments and so on.

(08:02) Shortly after the Coca-Cola ad came out, someone managed to replicate it in only 10 minutes, playing with the Pepsi brand, as you can see. This is, of course, not an official Pepsi ad. How was it done? On a platform that costs around 400 dollars per month. With 400 dollars per month, it is possible to replicate the Coca-Cola ad in 10 minutes, or better still, create an original ad from scratch for our own brand, using something very interesting: accumulated knowledge about what makes an ad successful in our business area. The platform is called Icon.

(08:57) Let us take one more example, perhaps even closer to us. I will show you an experiment we carried out at Liminal. This is a podcast where we curate a series of relevant books in the areas of marketing technologies, CRM, automation and related topics.

(09:21) With artificial intelligence, we have narrators, two people commenting on the books, summarising them and sharing the main insights in 15 or 20 minutes. I will show you a small part. CRM, BI, Liminal. Twenty minutes of summary and commentary between two hosts generated by artificial intelligence. How did we do this? On a platform that became viral a few months ago.

(10:06) Not many people know it yet, but it is incredible and free. It is Google's NotebookLM. What can we do with NotebookLM? We can do many things that we can already do with ChatGPT, which everyone knows, but with some very relevant details.

(10:27) We bring in the knowledge we want: books, videos, websites, anything, with dozens and dozens of sources as the knowledge base. Then we talk to that data, with the confidence that in this case the LLM will always reference the sources we provide and not some random external source.

(11:01) Among several very interesting features is Audio Overview. With a few prompts, we can make it generate audio focusing on the things we want the hosts to say or the observations we want artificial intelligence to make about those topics. Another example: this is a Liminal landing page. We talk about one of the platforms we work with, one of the brands, or simply a typical page with results, how we work and some important features.

(11:39) What I did a few days ago was this: I took this page, brought the link into ChatGPT and asked it to create an advertising image. I said, I want to make a magazine ad based on this information. It gave me copy suggestions and so on. I then said, no, I actually want this as an image.

(12:05) In a few seconds, this is what ChatGPT created. A few weeks ago, perhaps about a month ago, OpenAI rebuilt the image generation engine for the 4o model. And rebuilt it in a meaningful way. What we are seeing here are images very different from what artificial intelligence was producing until very recently: much more realistic, much closer to what we ask for, and with text.

(12:47) Yes, there are still some bugs and mistakes, but overnight we moved away from images where any text would almost always be wrong and the characters broken, and which were often not realistic for this type of image. And then, of course, perhaps you already know this, we can get here, as I did, and say: take this man's face, I want a female character. It generated the female character.

(13:12) Then I asked it to make the person look straight into the camera. All of this illustrates the first point I made earlier. First, these technologies are not only for brands with large budgets. Then there is a buzzword that is becoming increasingly popular: vibe marketing. What is this? I think, I am inspired, I have an idea, and I create.

(13:42) I do not need to wait for the designer, the person who knows Illustrator, the programmer or the person who understands HTML to quickly see my concept produced and test whether it is valid, or even, with some additional work, reach final-art quality. Where does this term vibe marketing come from? From a base term, very recent and increasingly used by programmers: vibe coding.

(14:14) Several very good programmers I speak with tell me this: Tiago, for weeks or months I have not written a single line of code. So this is not only about communication. Just as we can build communication actions with one prompt, or several good prompts, we can also create entire digital products. For example, using Replit or alternatives, we can conversationally create, or at least test, an idea for an app or software product, then take it and put it on the Google or Apple store.

(14:55) Second principle: be careful with what I have just done here. Be careful with the media fireworks that make us run after tools without first understanding two things, and a third that I will discuss later.

(15:16) First, what is this new reality of artificial intelligence? Worse than that, we often do not understand what type of artificial intelligence our business actually needs. The two stories I shared at the beginning, RedBalloon and Coca-Cola, illustrate the two major classes of artificial intelligence that currently exist.

(15:35) On one side, generative artificial intelligence: spectacular, creative, suddenly appearing, creating new worlds, keeping us awake at night, exciting us with the possibilities ahead, and frightening us with the possibilities of disruption to our business and profession.

(15:59) On the other side, analytical artificial intelligence, or what some call traditional or predictive AI: much more discreet, focused on analysing historical data, often requiring time series of historical data to produce something genuinely useful, but silently revolutionising our world for many years.

(16:24) This explains the rise of technological and economic phenomena such as Amazon, Google, Netflix and Facebook. Understanding these two broad categories of artificial intelligence implies knowing what we need to have prepared in our organisation to succeed in applying each one.

(16:49) For example, when we speak about analytical artificial intelligence, as we were saying a moment ago, Ana, we need to have basic data hygiene in place. What does this mean? First, a data-driven culture, making decisions not only based on instinct, which remains important, but also on information and data that help us make decisions. For that, we need quality data.

(17:21) In a word or two, how do we get quality data? Data governance. Who ensures that the information we have is useful? Who ensures that dark data, those unstructured data, or data that may even be structured, for example in Excel tables, but is not integrated or harmonised, is worked on so that it can be read and become useful for decision-making? We need information integration, understanding the difference between first-party data and third-party data, and reports with the key KPIs for real-time management.

(17:58) We cannot wait for a crucial report to take one, two or three days, or even weeks, which is what often happens. More than that, we need the capacity to take this information and these data and forecast the future.

(18:23) Predictive and prescriptive analytics. These are some of the fundamental conditions for taking advantage of analytical artificial intelligence. At the level of generative artificial intelligence, important topics include team training, capability building in prompt design or prompt engineering, how to create good prompts and how to test them, fine-tuning questions, and how to ensure that outputs are anchored in credible information that feeds the AI generation engine to produce useful results rather than noise.

(19:01) We also need to address topics such as grounding, red teaming, security and data encryption. What policies do we have? What can our employees paste into ChatGPT in the web version? Can they mirror everything from our organisation there? What type of data cannot go into web versions of large language models? When do we want to have a large language model installed on our own servers or computers to guarantee a certain level of information protection? These topics are very important.

(19:32) In truth, artificial intelligence is not one thing. It is a large umbrella under which everything fits. It is an abstract term, used mostly by marketing people to sell many things, but at its core it means this: every type of technological software application that mimics human behaviour.

(19:58) For that to happen, we need a large universe of probabilistic computing known as machine learning. We have all heard about machine learning algorithms. It is the combination of several algorithms that enables the results and use cases we have seen and will see shortly.

(20:22) The point is this: we need literacy to understand this new reality. In the same way that, 20 or 30 years ago, we needed to know the basics about the difference between hardware and software, an operating system and the applications running on it, or the difference between a mobile app and a web app.

(20:40) To make decisions about artificial intelligence today, we also need the literacy to understand, at least at a basic level, how different types of machine learning work, what limitations they have and what implications they may have for management decisions. I will give an example: deep learning algorithms. Deep learning algorithms are typically black boxes. It is very difficult to understand the criteria by which they made a specific decision.

(21:10) What does this mean? In critical sectors such as healthcare, banking and insurance, they cannot be used to produce certain types of results. Why? Because we need to be able to explain why a given loan was granted to one person and not another.

(21:33) Why did a person from a given ethnic background not gain access to credit? Was ethnicity one of the criteria that artificial intelligence used to make that decision? In marketing, the same applies. For customer profiling and automatic customer segmentation or clustering, what criteria were used to place a given customer in one segment and offer them a promotion, while not offering it to another?

(22:00) We need explainable AI, and deep learning algorithms are typically much more difficult to use for this. So this technology is not only for the rich, and we need to understand it in order to make good management decisions. The third principle is that none of this matters if we lose sight of what actually makes businesses exist: customers.

(22:25) If we continue the conversation focused only on technology, we are very poor. What we saw in recent decades, in the context of the fourth industrial revolution and digital transformation, was the use of technology with a focus on optimising processes, reducing waste and cutting costs.

(22:49) Today, that is no longer enough. It is possible to go much further. Look at the illustration with ATM machines. What happened in many cases was that we replaced the human touch, the presence of a human being, with the presence of a machine that, in practice, completely destroyed any possibility of warm, empathetic relationships with the customers on the other side. This is particularly worrying when we are talking about moments of truth, critical touchpoints in the customer experience and the customer's closeness to the brand.

(23:32) So where should we concentrate? On the customer experience and on customer expectations. We know that none of us wants to be bothered by brands that are only interested in selling on their own timing, what they want, and creating massive spam simply because they have the power to do it.

(24:03) We want a personal relationship with brands. We want brands to recognise us, remember us and be relevant. For that to happen, we need to know, in real time, the stage each customer or potential customer is at in their journey with us. This is also true in sectors where the decision cycle is short or more spontaneous.

(24:41) We need to understand what the customer wants to achieve at that stage of their decision journey or relationship with us. What is their goal? How can we help them achieve that goal and, ideally, help them better than the competition? This is about using technology to improve the customer experience.

(24:59) Let us move to the fourth and final principle. How can we design a marketing technology strategy that helps us with this? There are three pillars. This prevents us from having a fragmented logic, where we see a spectacular platform and buy it, then see another and buy it, and another and buy it.

(25:18) At some point, there is no strategy at all. Everything is loose and ad hoc. Where do we start? We can only improve the experience of each customer if we have a central nervous system that gives us a complete view of our relationship with that customer or potential customer.

(25:36) There is nothing new here. The software category is called CRM, Customer Relationship Management. It has existed for two or three decades. What I will show you now is what we should demand from a CRM today. How can artificial intelligence enter a platform like this and help us know our customers better and build closer, more empathetic relationships? I will give examples mostly focused on commercial prospecting and demand generation use cases in marketing.

(26:19) This is a CRM view, in this case HubSpot. We will stay with this platform today. What we have here is the record of a customer or potential customer. That is not the important point. Here is the person's name and data. I want to highlight two important points. The CRM must be connected to all our communication channels, at least the digital communication channels: website, newsletters, social networks and so on. This is one of the challenges many companies face when using CRM.

(26:49) It is often just a bureaucratic police platform for customer-facing staff, salespeople and support teams, who have to report things that happened, and little more. To overcome this, we must ensure that the CRM is connected at least to our digital channels, and at least to the employee's phone and email.

(27:16) Based on this, we can then understand whether an email sent to a customer was read or not. If we sent a video in that email, where did the person stop watching? Did they watch the whole video? Where did they show interest? On our website, have they been engaged? We launched a landing page or campaign. Did the person see it? How long were they on the page? What was their behaviour across our channels?

(27:42) The second aspect I want to illustrate is that there is nothing more strategic in marketing than segmentation. That is where everything begins. Who are our segments? It is not easy to answer this question with many CRM platforms, especially the more traditional ones. Even in HubSpot, it is not always easy, but as marketing leaders we have to demand it.

(28:15) Which persona does this customer or potential customer belong to? Based on that, I have an entire marketing approach, an entire marketing mix that must be adapted to that persona. It is not only communication; it is also the offers.

(28:40) Do I talk about these products or those variations? At what prices? I have an entire strategy around this. If I can identify the person on the other side and communicate with identified audiences, not just anonymous audiences, I should do it.

(29:00) That is one of the things a CRM platform should give us. Up to this point, there is no artificial intelligence yet. We will jump there now. Another thing the CRM should give us is what I am showing here. We need to know the journey stage, the relationship stage each customer is in.

(29:18) It should tell us exactly that. Are we looking at a customer, a returning customer, a customer at risk of leaving, a lead being worked by marketing, or a lead being worked by sales? These are very important pieces of information. Let us move to artificial intelligence.

(29:35) Imagine that a month or two ago I had a phone call, it could have been Zoom or Teams, with this person, a hypothetical name and case, Inês C. Now I have to follow up. I want to send her an email. What was the matter about? Have you made a decision? I no longer remember the details.

(29:57) How will artificial intelligence help me here? It will immediately summarise this conversation and tell me exactly what was discussed. Of course, I can later audit the whole call, but what I want specifically is to send a follow-up email and touch on the important points. So what do I do? I ask the copilot, in this case, to help me write that follow-up email.

(30:21) This is a prompt I had already prepared. I ask it to address the concerns raised by the hypothetical client, such as integration issues and so on. I also want the email to mention a video or webinar we have on this topic.

(30:44) I have the prompt ready. It helps me produce the email. From there, I can take it into email, send it directly or make changes. The important point is this: I did not have to copy this recording and paste it into ChatGPT. In fact, perhaps I should not. In this case it is fine, because it is hypothetical, but in a real case perhaps I should not.

(31:11) This is proprietary information and must stay in my system. More than that, beyond this information and this call, there are many other interactions that this person has had with me and with my company, which artificial intelligence would not otherwise know. In this case, it does. It knows the context and knows whether this person has been a customer in the past or not.

(31:31) So it helps us read all this context and adapt, personalise, this very simple touchpoint: an email. Let us look at another example. Imagine I want to launch a campaign to acquire new customers. Here, I am in the business-to-business market.

(31:56) What I want is to take my segmentation criteria and ask artificial intelligence to help me identify companies, in this case relatively recent software companies with more than 100 employees and based in Portugal, companies that fit this segment. It thinks for a moment and suddenly we have a list with suggestions.

(32:33) Let us expand this to see it better. One company draws attention: Feedzai, one of the Portuguese unicorns and, incidentally, our neighbour at Instituto Pedro Nunes in Coimbra. I can ask for more information about Feedzai. What technologies do they use that are detectable on their website, for example? What is their size? What are the direct links to LinkedIn?

(32:59) I can say: when was Feedzai's last funding round? Is this a good moment, from a cash-flow perspective, for them? Who are Feedzai's main competitors? You can see funding rounds, a last one listed in 2015 and more recent ones in 2021, plus main competitors.

(33:18) Based on this information, I can create an action, in this case a one-to-one prospecting action, highly personalised. I quickly understood what I needed to understand about the company's context, or I could create a more massified action originating in marketing to reach this company.

(33:42) Now the question is: is this a good moment to speak with them? Imagine Liminal's case. We work with CRM systems such as Salesforce and HubSpot, and with automation systems. Is this a good moment to address this topic with Feedzai? What we would like to have is an alert when, inside those companies, someone starts searching online for these terms and topics that matter to us.

(34:07) Why? So we can interact with them at the moment when these topics are probably more interesting and relevant. What I am showing here is a HubSpot resource, and many other platforms also have this, called intent-based signals.

(34:41) What is this? We come here and select a series of keywords, for example Salesforce partners. HubSpot and artificial intelligence monitor several points online and show us searches taking place within different companies around these terms: Salesforce Partners, certified and so on. A thousand companies are searching for this term. I added it to my topics, and I could add many others.

(35:04) What does this then show me? A list of those companies. This is only B2B. Other platforms are a little more invasive and also show contacts. In this case, they do not. But you can see a list of hundreds, thousands, even 44,000 companies. Are they all relevant to me? No, because I have a segmentation strategy.

(35:35) So I can go to my CRM and say: bring me only companies that match my target segments. Simplifying, all companies in Portugal. HubSpot changes the list. One, two, three companies that are currently searching online for the terms I defined. I can then filter for companies that are not yet in my CRM, that are already in my CRM but are not yet customers, and so on. And of course, this works not only for the terms I mentioned, but also for fintech, payroll processing and many other areas.

(36:05) We are nearly at the end, but there are still two important pillars I need to share with you. CRM is the central nervous system that helps us understand all this context and the right moment to speak to the people on the other side. But we need to reach them.

(36:30) For that, we need automation platforms. Not every CRM gives us this automation capability. When we talk about automation, we are talking, in simple terms, about creating campaigns that run around one person at the right moment, with the right message, through the right channels. At minimum, digital channels.

(36:53) This is the simplest concept, and it does not necessarily involve artificial intelligence. At the more sophisticated or advanced frontier of automation today, we have AI agents. What I am going to show you is a prospecting agent.

(37:15) The goal of this artificial intelligence agent is to interact autonomously or semi-autonomously with the accounts and people I have selected, book meetings and eventually pass those leads to marketing or sales to close business.

(37:35) There are other agents. HubSpot, Salesforce and Microsoft have been working a lot on this, enabling customers to access prospecting agents, marketing agents that create campaigns, customer support agents and so on. Example of a prospecting agent: what do we do? We train it.

(37:51) We tell it: this is my company, this is what differentiates me. I solve these problems, these are my products and services, this is my segment, these are the goals. My goal is for you to book a meeting. Here is some guidance for touchpoints in emails, how you should speak, what tone of voice to use, and so on.

(38:11) I brought in another illustrative dummy contact, John D., who works at Adidas. Let us see the result of this prospecting agent with this context. What did it do? It read all the information about John D. in the CRM and all the information about the company. It looked for a series of data points and is already suggesting an approach in three emails, separated by two days, to achieve the goal I defined, which was booking a meeting.

(38:50) It says: hello John, with the launch of the Adidas Golf collection, it is natural that pressure is increasing on marketing, sales and operations teams. At Liminal, we specialise in solving this type of problem with artificial intelligence, CRM, automation and so on. Very interesting. This can work autonomously or semi-autonomously.

(39:17) We are seeing it because it is in a semi-autonomous format. But there is an important point: what about hallucinations? How can we be sure that there is no invention in this text, that generative artificial intelligence is not setting a trap for us and that the effect on the other side will not be much worse, as is obvious?

(39:38) We can, as I mentioned earlier when speaking about explainable AI, check all the sources of information. What does the CRM know, what does artificial intelligence know about this person and this company, and what did it use to produce this content?

(39:58) In the middle, we have a recent article. When I created and called this person into the process, this article had been published only a few hours earlier. That was the source that inspired the approach and contextualised the entire message. Very well. Final step: we spoke about CRM, and we spoke about automation to reach the right person at the right time.

(40:24) Now we need to verify whether what we are doing worked or not. For that, we need a third technological pillar in our marketing stacks: business intelligence platforms such as Power BI and Tableau. There are many alternatives, but these are among the most relevant.

(40:45) A business intelligence platform should show us our business by the day, hour or second, with a single click. In marketing, for example, we are investing in organic search, optimising our website to appear in the first Google results. We are investing in email marketing, social media, paid media and so on.

(41:09) We want to understand what impact that is having, yes, at the level of the metrics relevant to those channels, but then we must get out of vanity metrics and understand to what extent the investment we are making, for example in organic search, is or is not bringing customers and return. This type of information is critical for businesses, and it is critical that it is available in real time.

(41:34) Retrospective information matters, but what else do we need? Based on these data and others, we need to draw predictive conclusions about what is likely to happen in the future. At least that. The example I will give is artificial intelligence helping us with a specific deal.

(41:58) The case is this: Mr. Talomali wants a week of holiday in Paris and has plenty of budget because he will spend 7,000 euros on it. Great. What is the probability that I will close this deal? Artificial intelligence is making that prediction here, and it does not look good. Fourteen points on a scale that I believe goes to 100, and those 14 points have been decreasing recently.

(42:36) What is artificial intelligence seeing? It is looking at a set of data points and saying: there has been no response from the other side to this email for a while, no one has followed up for a while. Those are bad signs. Comparing this deal and this customer's profile with other successful deals, the outlook is not positive.

(43:09) Based on this, we can use data to understand our reality, forecast the future and follow a systematic logic when approaching artificial intelligence in our businesses, much more than an ad hoc or random logic. This is what we have been doing with several of our clients, such as Centro Cultural de Belém and GoContact.

(43:28) One resource I would like to leave with you is a free diagnostic, available on our website, to understand the maturity level of your marketing organisations or brands in relation to these topics of automation and marketing personalisation.

(43:28) The challenge is exactly this: intelligence is nothing more than the ability to adapt to change. And change today requires companies with digital brains, but brands that are increasingly human.

(43:28) Thank you very much, Tiago, for your excellent presentation. Before passing to the audience, I would like to ask a question, because we are clearly living through a period of technological disruption. Never before have we had so much technology at our disposal, and the temptation is great. As you explained, this means enormous gains in productivity and also in sales conversion.

(44:10) But at the same time, consumers increasingly scrutinise brands for authenticity and transparency. How, in your opinion, can we reconcile these two dimensions? Absolutely. There is a critical difference between marketing oriented towards growth and, when we talk about automation, that association is always present, and marketing oriented towards development.

(44:36) What do I mean? Marketing focused on scaling results quickly, often aggressively, the spammer logic: more leads, more clicks, more conversions. Today's technologies can give marketers almost Big Brother-like power. However, just because we can does not mean we should.

(45:03) The future of marketing is not really about knowing more about the consumer, nor about having greater ability to shout louder and louder than competitors. It is about knowing how to take better care of the relationship with the consumer, with the customer and even with society.

(45:33) Strategic intent determines the impact we create with technology, or the impact technology has. The truth always comes to the surface, and the market notices it. But this has to start as a strategic decision. I do not want merely to invoice and sell at all costs because I have a target to hit or because otherwise I will be fired. No.

(45:57) The mindset of management teams and company boards has to change from a mindset of exploitation to a mindset of development. That means recognising that on the other side there is not just a customer, but a human being. There is not just a country, but a society, and we are part of it.

(46:18) If this mindset exists, then when we take these platforms, we take them with integrity, or we must take them with integrity. Of course. And that is also where human supervision is important, to guarantee in some way that the brand remains authentic.

(46:51) I was also hearing about customer experience management and emotional peaks, or experience peaks. There is an author who speaks about peak-end experiences, and consumers do not remember the experience sequentially, but rather the moments that had the greatest impact. From your experience, what can generate the greatest impact on a consumer? What can create an experience peak?

(47:14) That is an important topic. First, whenever we are surprised, that is memorable. And often that surprise requires no technology at all. It is enough to be in the right place, at the right time, doing the right thing, which means knowing the customer and having a business model suited to that. Creativity also plays a role, as it always has. How can creativity and communication surprise, or genuinely deliver much more at the right moments than the customer expects or the competition offers?

(47:50) There is nothing wrong with having moments in the customer journey where the customer says: you are not very good at this, the competition is much better. There is no problem with that, as long as it was a deliberate choice made by the brand.

(48:08) In these points, I do not want to be extraordinary, I do not want to be better than the competition. In those other points, I will be extraordinary and I will surprise. There are two other points. We do not always have to exceed customer expectations.

(48:38) When we talk about Customer Effort Score and the studies around it, often what we need to do is not go the extra mile, but simply solve quickly and well whatever problem the customer has. There is a false glorification around extraordinary experiences that are far above customer expectations.

(48:58) Finally, hyperpersonalisation is usually a wow moment, partly because we are still not used to it. But there is a danger with hyperpersonalisation: these people know too much about me, and I do not want that. So here too we must know where to take hyperpersonalisation, which points in the journey should be hyperpersonalised, and where we should simply step back and leave the person in anonymity.

(49:24) Very well. Empathy above all, right? Yes, completely. I will now pass the floor to the audience. Does anyone have a question before I move on to the other topics I had here? Diogo, good evening. I saw you there. Hello. Good evening. Can you hear me properly? Perfectly. Very well. I have two questions.

(49:51) First, thank you. It has been a very good learning experience listening to Tiago. Tiago mentioned hallucinations earlier. That made me think that many of the agents he presented can be used in dozens and dozens of cases. One of the ways he referred to, and correct me if I am wrong, was human control or supervision, where we could verify where the information was obtained and how.

(50:25) How can we control that with dozens and dozens of agents working at the same time? It becomes almost unlikely that I can track everything. That is one question. The other question that made me curious, now that you were discussing it, concerns hyperpersonalisation.

(50:45) As we know and learn at IPAM, concern about data is increasingly constant. However, we also know that by segment, or from segment to segment, that concern can be completely different. How do we manage hyperpersonalisation, and how far should we go? What is our limit according to each segment? How do we achieve that dexterity, given that we often work with dozens and dozens of different segments within the same audience?

(51:11) Those are my questions. Thank you very much. Excellent questions, Diogo. I do not even know where to start. To simplify and also respect time, on hyperpersonalisation, a good recommendation would be to think about the concept of zero-party data. Let us ask people. Let us speak with people. Let us understand what they want and how they want their experience to be managed by us.

(51:43) To what extent can we be a little more intrusive or more persistent? Yes or no? We need to understand that from the other side. Then, of course, we must test. In which segments do we get good results when we personalise or hyperpersonalise? When do we get poor results? What is the prediction around that? With some platforms, not HubSpot, but other customer journey orchestration platforms, we can understand whether certain segments react better or worse to WhatsApp, SMS, email marketing and so on, and make decisions accordingly.

(52:14) To be succinct, the other question concerns hallucinations. So we automate, automate, automate, and at some point we need a human being watching all the automations, otherwise things go wrong.

(52:42) These types of projects should usually be carried out first as a pilot, with very specific use cases where, at the beginning, there is meticulous supervision. When we think about bringing artificial intelligence into our businesses, we need to think of it as bringing in a super-intelligent recent graduate, very convinced that they know everything, but who we know still falls short in many areas.

(53:18) Therefore, we need to stand beside it and give it specific, routine tasks that we already know how to do well, tasks that we have already tested manually or otherwise. Then we verify whether the results are consistent. Once they are consistent, we can broaden, broaden and broaden the scope, and eventually use sampling methods to monitor and correct deviations.

(53:38) Interesting. As I was listening to you, I immediately remembered a well-known gym brand, which I will not name here. I was a customer, although I have been a customer of almost every gym in Porto. One of the things they ask at the beginning is: how do you want to be contacted? I immediately said email, because I value my privacy. Then they kept phoning me several times. So that is really not knowing the customer, and really disrespecting the customer's will.

(53:58) Let us hear Fernando. Good evening, or still good afternoon, Fernando. Good afternoon. Can you hear me? Yes. Thank you for the presentation. It was very interesting and it relates to the presentation itself and to the CRM or system used. All the applied examples were examples inserted, automated and integrated in the CRM, in your CRM.

(54:17) The platform I showed is called HubSpot. There are others. But is that platform outside your CRM, or does it only work integrated with the CRM? Let me explain. The question concerns another company, another client, that has another CRM. Can that integration be done, or is some middleware or another system needed to make the integration?

(55:19) Excellent question. First, we work with several brands and software houses, namely HubSpot, which was the system I showed here, among others. Salesforce also produces this type of technology, Microsoft does too, Zoho also produces this type of technology, and Amplemarket as well.

(55:40) Why did I choose to show you HubSpot? Because it is probably one of the best solutions for solving exactly the issues Fernando mentioned: integration. I have several applications and now I want all of them to talk to each other easily. Why? Because the business model is all-in-one. For example, when I spoke about business intelligence platforms, I mentioned Power BI and Tableau, which dominate the market, but then I showed you HubSpot again.

(56:06) Why? Because it also has a business intelligence solution inside its application, which means that whatever we do, and whatever HubSpot products we buy, the databases work well together, everything is highly integrated and the user experience is the same. That does not happen with all brands.

(56:32) Now, the final point is this. What do we need? Best-of-breed or all-in-one is one of the typical IT decisions for marketing areas. Best-of-breed means: HubSpot may be interesting for prospecting agents, as I showed, but wait, because HubSpot's Portuguese is still not that good. Perhaps we need a platform like Amplemarket, which even synthesises our voice and can send synthesised voicemail messages, moving beyond the email marketing channel.

(57:12) Another example is intent-based signals. HubSpot is now coming to market with this solution, understanding who is looking for information related to the campaign we want to launch, the product we have, or whatever it may be. There are other platforms that do only this. When we look at best-of-breed, we know, first, that we will certainly pay more, and second, that it will often be much better than what is already included in all-in-one platforms.

(57:38) Finally, we need to ensure how the integration will work. We do not want to hire a developer to write code and integrate APIs. We need to understand whether native integrations already exist, what limitations they have, and whether, if there are no native integrations with our CRM, which will be the central platform for all of this, we can use an integration platform as a service, such as Zapier or n8n, to connect these platforms simply.

(58:15) Very well. Thank you for your questions. I do not know if anyone else has a question. Hugo, are you still there? Yes, absolutely. I was enjoying it a lot. First of all, congratulations on the initiative. I think Portugal needs more discussions like this. European companies need to embrace AI once and for all and understand that the change we are going through is unprecedented.

(58:46) Congratulations also on the master's programme. It arrives at the right time. We truly need these skills. Our executives, our recent graduates and our experts need to adapt to this new reality. And congratulations to Liminal too, because it is a beacon in the country for moving forward on this front.

(59:08) We need more companies like Liminal that push this frontier. There will be a lot of work over the next 10 years to embrace and manage all the change required for this new reality. In trying to please those we serve, in trying to exceed expectations and differentiate, the complexity of maintaining competitiveness will be increasingly unprecedented. We will all have to be much stronger.

(59:43) Our small and medium-sized companies will have to be able to compete with the stacks that are being silently built. I predict that agents will begin not only to suggest what we buy and what we find, but also to execute on our behalf and even buy on our behalf.

(1:00:12) Fantastic opportunities are now emerging with new protocols and new methodologies to make our products and services more discoverable. Just as a small improvement, I think that within these three vectors, the hyperpersonalisation discourse is missing a critical ingredient: knowledge of the product and service portfolio.

(1:00:32) In other words, if we want to hyperpersonalise, we need to know a great deal about our customer and what they do, but we also need to know a great deal about the metadata of our products and services, so that we can combine both sides. Congratulations. I think we need more sessions like this.

(1:00:52) Professor, please hold more sessions like this so that we can continue this discussion and all learn. Of course. Thank you, Hugo, for your contribution. I was going to finish by asking Tiago how he sees the marketing professional of the future, because there is also talk that human skills will become increasingly important. I would like to hear your opinion.

(1:01:12) How do you see the marketing professional of the future? It is the question I ask myself every day: where will I be in five or 10 years? A study came out a few months ago, I believe by McKinsey in Portugal, on the impact of artificial intelligence on jobs, and it was somewhat frightening: more than one million jobs affected.

(1:01:36) Then they used those terms, those euphemisms, reskilling and so on. One of the areas most affected, particularly by generative artificial intelligence, is precisely marketing, followed by accounting. That same study also spoke about the importance of professionals gaining more technological skills.

(1:02:06) Technological skills here relate to what I mentioned earlier: at least literacy. We have to become increasingly technological, because today speaking about software and working with software does not necessarily mean knowing how to code or develop lines of code. At the same time, the study was very clear that soft skills will be decisive, especially our ability to relate to the human beings on the other side. I think that the more digital brands become, the more those that have the human being present at the right moment will differentiate themselves.

(1:02:41) Critical thinking, ethical judgement. We press a button in ChatGPT and it does all our homework, answers exams and so on. But to what extent is that correct? Is it not? Are we going to accept it? Creativity itself is also critical.

(1:03:03) Artificial intelligence, and this is a major debate, but I do not agree that artificial intelligence is creative. It does not create anything truly from scratch. Where is originality? Where do knowledge and fields of expertise intersect? Vision, insight, individual experience, empathy and interpersonal communication are all increasingly relevant attributes.

(1:03:27) The ability to create and maintain relationships will be decisive. Yes, yes. I am one of those people who thinks there will also be opportunities for businesses fundamentally based on human experience, because it will indeed become scarce, and scarcity creates desire.

(1:03:52) Perhaps people will pay a premium for those services. I imagine this, for example, in luxury brands. Luxury brands really have to surprise consumers, as you mentioned earlier, and exceed expectations. So those brands will probably still need to be strongly based on human experience. I am just thinking out loud.

(1:04:11) I do not know whether anyone else would like to ask a question. We are here today within the scope of the master's degree in Marketing and Technology, a unique master's degree in Portugal lasting one year, which promotes the development of technological, marketing and analytical skills that are now fundamental.

(1:04:38) Tiago, all that remains is to thank you very much. It was excellent, and it is a pleasure to have you on the master's programme. We could spend the rest of the night debating these topics, which are so important and, as Hugo also mentioned, so relevant for Portuguese companies and beyond, because the challenges are many and we still do not have all the answers.

(1:05:13) We have the questions, as Peter Drucker would say, and it is from these debates that light emerges. In essence, we are all searching for the purpose of brands and the future of brands, for how the experience brands will provide to consumers in the future will really look. Technology has truly become embedded in our lives, and it is impossible to avoid these topics.

(1:05:32) Well, if there are no more questions, I want to thank everyone for your presence and make myself available for whatever you may need. I also remind you that information about the master's degree in Marketing and Technology is available on the IPAM website. We are available to clarify any questions. Thank you very much once again to everyone, and have a good evening. Thank you very much. Thank you. A hug to everyone.

(1:05:51) See you next time. Thank you very much. Thank you.