WEBVTT 1 00:00:00.359 --> 00:00:03.159 Hello, welcome to the XMACNA podcast. 2 00:00:03.159 --> 00:00:05.600 I'm Gustavo Fechus, and today 3 00:00:05.600 --> 00:00:06.879 we are here once again to 4 00:00:06.879 --> 00:00:09.440 talk with Lucca Carvalho and 5 00:00:09.440 --> 00:00:12.519 Rômulo Carneiro. Well, Lucca, Rômulo, we 6 00:00:12.519 --> 00:00:14.440 have some important issues to 7 00:00:14.440 --> 00:00:17.000 discuss here. I 8 00:00:17.000 --> 00:00:20.600 wanted to start by saying the following, and 9 00:00:20.600 --> 00:00:24.680 we, in interactions with the market, have already 10 00:00:24.680 --> 00:00:28.119 noticed a certain confusion, Lucca, about 11 00:00:28.119 --> 00:00:31.000 the fact that chatbots already exist 12 00:00:31.000 --> 00:00:33.200 out there, right? I mean, I think 13 00:00:33.200 --> 00:00:36.360 unfortunately, who hasn't 14 00:00:36.360 --> 00:00:38.079 had an experience, let's say, 15 00:00:38.079 --> 00:00:41.000 horrible or terrible with a chatbot that 16 00:00:41.000 --> 00:00:43.480 cast the first stone. But what I 17 00:00:43.480 --> 00:00:46.120 wanted to point out, Rômulo and Lucca, is the 18 00:00:46.120 --> 00:00:48.960 following: XMACNA aims to 19 00:00:48.960 --> 00:00:50.960 develop the concept of a digital 20 00:00:50.960 --> 00:00:53.000 employee in its various 21 00:00:53.000 --> 00:00:56.559 applications. And I think that in no way should it 22 00:00:56.559 --> 00:00:58.120 um 23 00:00:58.120 --> 00:01:01.760 be confused with the idea of a chatbot, right? 24 00:01:01.760 --> 00:01:04.360 So let's start by making that distinction 25 00:01:04.360 --> 00:01:06.640 first, so that, who knows, with 26 00:01:06.640 --> 00:01:09.159 some luck, we’ll never again 27 00:01:09.159 --> 00:01:11.560 get this question. And then we 28 00:01:11.560 --> 00:01:13.439 can move on later to other 29 00:01:13.439 --> 00:01:15.080 parts of this story, at least if 30 00:01:15.080 --> 00:01:16.360 someone asks, we’ll have this clip ready. 31 00:01:16.360 --> 00:01:18.920 Send this, send this exact link. 32 00:01:18.920 --> 00:01:21.680 Alright, go ahead. So Lucca, right? Well, I think... 33 00:01:21.680 --> 00:01:23.880 This answer, if you allow me to take two... 34 00:01:23.880 --> 00:01:26.280 Steps back. By the way, thank you... 35 00:01:26.280 --> 00:01:27.720 For the audience, thank you for being here... 36 00:01:27.720 --> 00:01:29.920 With us today. Thank you for the invitation. 37 00:01:29.920 --> 00:01:32.720 Thank you, Rômulo, for being here with us. Well... 38 00:01:32.720 --> 00:01:35.439 XMACNA was born within the Gir network... 39 00:01:35.439 --> 00:01:36.840 Right? It started as a spinoff from a... 40 00:01:36.840 --> 00:01:39.520 Education company, and in the Gir network we... 41 00:01:39.520 --> 00:01:41.000 We were already working with artificial intelligence... 42 00:01:41.000 --> 00:01:44.960 AI, we had been doing research for years, and... 43 00:01:44.960 --> 00:01:47.479 To help with our core activity... 44 00:01:47.479 --> 00:01:50.759 Which is correcting essays, teaching... 45 00:01:50.759 --> 00:01:53.079 Students to write better. 46 00:01:53.079 --> 00:01:56.320 Well, the first digital employee... 47 00:01:56.320 --> 00:01:58.119 Of XMACNA... 48 00:01:58.119 --> 00:02:00.719 Who still works, it was for the Gir network and I... 49 00:02:00.719 --> 00:02:02.079 Actually, I think our employee... 50 00:02:02.079 --> 00:02:05.280 Of the year some years ago was Bárbara, who... 51 00:02:05.280 --> 00:02:06.799 Is not a chatbot, it has nothing to do with... 52 00:02:06.799 --> 00:02:08.759 A chatbot. When we started to... 53 00:02:08.759 --> 00:02:10.000 Work with these artificial intelligences... 54 00:02:10.000 --> 00:02:11.239 When we started to work with language models... 55 00:02:11.239 --> 00:02:12.920 To work with language models, we... 56 00:02:12.920 --> 00:02:14.720 Realized that we would be able to... 57 00:02:14.720 --> 00:02:16.760 Automate some cognitive processes... 58 00:02:16.760 --> 00:02:19.319 Inside companies, eliminating a step... 59 00:02:19.319 --> 00:02:20.800 That before had to be done by some... 60 00:02:20.800 --> 00:02:23.080 Human being. Usually, it's a step... 61 00:02:23.080 --> 00:02:24.599 It's very boring, and a step that humans 62 00:02:24.599 --> 00:02:26.400 didn't want to do, freeing up the human 63 00:02:26.400 --> 00:02:27.720 to be able to do more 64 00:02:27.720 --> 00:02:29.360 interesting things. Give an example of what 65 00:02:29.360 --> 00:02:32.200 you're talking about. In the case of Redigir, 66 00:02:32.200 --> 00:02:34.879 Bárbara, she works by doing the 67 00:02:34.879 --> 00:02:37.040 guidance note, the final comment 68 00:02:37.040 --> 00:02:39.360 on the essays, which is extra work for the 69 00:02:39.360 --> 00:02:42.040 grader. The human grader, flesh and blood, 70 00:02:42.040 --> 00:02:43.560 has to read a text and make a 71 00:02:43.560 --> 00:02:45.560 series of annotations and then assign a grade. 72 00:02:45.560 --> 00:02:47.239 After that he has to do a 73 00:02:47.239 --> 00:02:49.440 review of everything he marked and 74 00:02:49.440 --> 00:02:51.599 explain in Portuguese to the student how 75 00:02:51.599 --> 00:02:53.640 they have to, how they can correct 76 00:02:53.640 --> 00:02:56.000 that mistake or those 77 00:02:56.000 --> 00:02:58.840 mistakes. What happens is our 78 00:02:58.840 --> 00:03:01.319 graders do this at scale. We 79 00:03:01.319 --> 00:03:04.400 grade thousands of essays a day. A, a 80 00:03:04.400 --> 00:03:06.280 professional grader who's there 81 00:03:06.280 --> 00:03:08.920 at the front line, the first essay they grade with 82 00:03:08.920 --> 00:03:11.640 standards, the second with a certain 83 00:03:11.640 --> 00:03:13.400 final comment. It decreases a bit by 84 00:03:13.400 --> 00:03:16.799 the 12th, the 30th, and that keeps getting a little 85 00:03:16.799 --> 00:03:18.959 lazier, because patience runs out, right? 86 00:03:18.959 --> 00:03:23.560 We created a copilot that comes in 87 00:03:23.560 --> 00:03:26.319 right at this part, taking away the 88 00:03:26.319 --> 00:03:28.239 boring part of the grader's work 89 00:03:28.239 --> 00:03:30.120 so the human grader gains 90 00:03:30.120 --> 00:03:33.159 speed, quality, and quality of life. 91 00:03:33.159 --> 00:03:35.360 time and earning more money for the 92 00:03:35.360 --> 00:03:36.720 same type of work. In fact, 93 00:03:36.720 --> 00:03:38.480 we remove a boring part of the job 94 00:03:38.480 --> 00:03:40.680 for them. They can be more productive and still 95 00:03:40.680 --> 00:03:42.319 earn more for the same amount of time worked. 96 00:03:42.319 --> 00:03:45.120 This was a turning point within 97 00:03:45.120 --> 00:03:48.480 XMACNA and we realized that, well, 98 00:03:48.480 --> 00:03:50.480 just as XMACNA had a Bárbara, 99 00:03:50.480 --> 00:03:52.000 there was actually room for several 100 00:03:52.000 --> 00:03:53.760 Bárbaras inside. Then others started appearing, 101 00:03:53.760 --> 00:03:55.959 like the Oracle, who is a programmer, 102 00:03:55.959 --> 00:03:57.840 he makes requests to our database 103 00:03:57.840 --> 00:03:59.760 using natural language, for example. 104 00:03:59.760 --> 00:04:01.799 Then there’s Keber, who is our 105 00:04:01.799 --> 00:04:04.120 data scientist, who analyzes the 106 00:04:04.120 --> 00:04:05.439 performance of a school and makes a 107 00:04:05.439 --> 00:04:07.200 suggestion for a pedagogical intervention, 108 00:04:07.200 --> 00:04:09.159 which is a suggestion for a teacher 109 00:04:09.159 --> 00:04:10.560 to apply in their classroom. 110 00:04:10.560 --> 00:04:13.720 We realized that every type of 111 00:04:13.720 --> 00:04:15.799 company could have Bárbaras, could have 112 00:04:15.799 --> 00:04:17.560 cognitive processes that could be 113 00:04:17.560 --> 00:04:20.959 automated and, while the market 114 00:04:20.959 --> 00:04:22.479 later, we realized what 115 00:04:22.479 --> 00:04:23.880 was going to happen. It’s natural that 116 00:04:23.880 --> 00:04:26.320 it happened. People or companies that 117 00:04:26.320 --> 00:04:28.880 worked with chatbots saw in 118 00:04:28.880 --> 00:04:31.160 artificial intelligence a great 119 00:04:31.160 --> 00:04:35.720 ally for chatbots, and it is well suited 120 00:04:35.720 --> 00:04:37.680 for this role. Without 121 00:04:37.680 --> 00:04:38.440 Problem. 122 00:04:38.440 --> 00:04:42.160 Yeah, but there are a lot of Frankensteins out there, right? 123 00:04:42.160 --> 00:04:43.759 People who worked at companies, 124 00:04:43.759 --> 00:04:46.800 who worked with... had an entire 125 00:04:46.800 --> 00:04:48.240 system set up to handle the 126 00:04:48.240 --> 00:04:51.479 chatbot. The chatbot, it has a... it 127 00:04:51.479 --> 00:04:54.160 assumes a very rigid flow. It 128 00:04:54.160 --> 00:04:56.880 assumes knowing the questions and the 129 00:04:56.880 --> 00:04:58.639 answers. Well, no need to explain here, 130 00:04:58.639 --> 00:05:00.759 right? Everyone has already had a chance 131 00:05:00.759 --> 00:05:03.440 to get frustrated with a chatbot. And a 132 00:05:03.440 --> 00:05:05.400 digital employee, when placed 133 00:05:05.400 --> 00:05:08.600 in this sales position, it revolutionizes. 134 00:05:08.600 --> 00:05:11.000 It's something completely different, it's 135 00:05:11.000 --> 00:05:12.720 a different experience, right? A different experience. 136 00:05:12.720 --> 00:05:15.160 But it doesn't stop there, because the process 137 00:05:15.160 --> 00:05:18.160 the cognitive process involved there 138 00:05:18.160 --> 00:05:21.080 is that the salesperson has to perform, not 139 00:05:21.080 --> 00:05:24.160 it's not just about talking, it's not limited to 140 00:05:24.160 --> 00:05:26.000 being there interacting with the public 141 00:05:26.000 --> 00:05:28.160 there are a number of other 142 00:05:28.160 --> 00:05:30.319 back office activities that are 143 00:05:30.319 --> 00:05:32.919 repetitive, that take a lot of time, that 144 00:05:32.919 --> 00:05:35.160 no salesperson likes to do, and that 145 00:05:35.160 --> 00:05:36.720 our artificial intelligences also 146 00:05:36.720 --> 00:05:40.120 are involved and optimizing, saving 147 00:05:40.120 --> 00:05:42.160 the salesperson's life, right? We say that the 148 00:05:42.160 --> 00:05:45.600 CRM is the main ally and the 149 00:05:45.600 --> 00:05:47.639 main enemy of the salesperson, because when 150 00:05:47.639 --> 00:05:50.479 filled out properly, it's, it's, it's a miracle 151 00:05:50.479 --> 00:05:53.160 If it's filled out wrong, it sabotages you. So the 152 00:05:53.160 --> 00:05:55.840 our solution, the solutions from XMACNA 153 00:05:55.840 --> 00:05:57.039 they don't start out like the 154 00:05:57.039 --> 00:05:59.639 Frankenstein of some other system, 155 00:05:59.639 --> 00:06:00.680 a system that later adds 156 00:06:00.680 --> 00:06:02.360 artificial intelligence in the middle. It 157 00:06:02.360 --> 00:06:06.080 is AI-first from the start. So we 158 00:06:06.080 --> 00:06:09.199 designed the whole process, thought through every 159 00:06:09.199 --> 00:06:11.400 step, understanding the 160 00:06:11.400 --> 00:06:12.639 potential of Artificial 161 00:06:12.639 --> 00:06:14.759 Intelligence, and we created something that, well, 162 00:06:14.759 --> 00:06:17.720 doesn't exist in the market. Which, well, I think 163 00:06:17.720 --> 00:06:20.000 I'll hand it over to Rômulo because he's 164 00:06:20.000 --> 00:06:22.479 the one who leads the factory and is there with 165 00:06:22.479 --> 00:06:24.560 an army of Autobots. We call the 166 00:06:24.560 --> 00:06:26.039 Autobots our digital employees 167 00:06:26.039 --> 00:06:27.800 who are in the back office. He's 168 00:06:27.800 --> 00:06:29.160 there configuring these Autobots. You can 169 00:06:29.160 --> 00:06:31.520 tell us about the revolution it is in a 170 00:06:31.520 --> 00:06:33.800 company that needs to serve at 171 00:06:33.800 --> 00:06:35.919 scale. So, how is it, Romulo? Of course, 172 00:06:35.919 --> 00:06:38.360 guys, here's the thing: taking a step 173 00:06:38.360 --> 00:06:40.280 back in this point that Lucca 174 00:06:40.280 --> 00:06:42.319 was bringing up, I think it's interesting 175 00:06:42.319 --> 00:06:44.720 for us to understand a difference of ours at 176 00:06:44.720 --> 00:06:48.000 XMACNA compared to chatbots, even 177 00:06:48.000 --> 00:06:51.000 chatbots that work with AI today, is that 178 00:06:51.000 --> 00:06:53.840 the idea at XMACNA always starts in 179 00:06:53.840 --> 00:06:56.160 cognitive processes. So I think the 180 00:06:56.160 --> 00:06:58.599 first stage of our work with 181 00:06:58.599 --> 00:07:00.319 company, which we call the meeting of 182 00:07:00.319 --> 00:07:02.599 KickOff, that initial meeting, is to say 183 00:07:02.599 --> 00:07:04.280 like this: Gustavo, you who are 184 00:07:04.280 --> 00:07:06.720 hiring us, right? What is the process 185 00:07:06.720 --> 00:07:08.400 that you want to automate within your 186 00:07:08.400 --> 00:07:10.960 company? What is your complete workflow like 187 00:07:10.960 --> 00:07:13.960 around this stage of the process, where 188 00:07:13.960 --> 00:07:15.680 your fun... where the digital employee 189 00:07:15.680 --> 00:07:17.639 will act? How does he 190 00:07:17.639 --> 00:07:20.080 process within your entire funnel? So, 191 00:07:20.080 --> 00:07:23.639 there is a very thorough analysis 192 00:07:23.639 --> 00:07:26.440 individualized of what the steps are 193 00:07:26.440 --> 00:07:28.960 of your company specifically. So, it's not 194 00:07:28.960 --> 00:07:30.800 something that will just talk about 195 00:07:30.800 --> 00:07:33.240 the generalist knowledge that AIs 196 00:07:33.240 --> 00:07:35.800 often have. So, there is a cut. 197 00:07:35.800 --> 00:07:38.120 There are layers that are applied 198 00:07:38.120 --> 00:07:39.720 within artificial intelligence to 199 00:07:39.720 --> 00:07:42.120 make it extremely assertive 200 00:07:42.120 --> 00:07:44.319 within your workflow. I have already participated in 201 00:07:44.319 --> 00:07:47.479 some of these meetings a few times and 202 00:07:47.479 --> 00:07:50.400 there's something cool, Lucca, that 203 00:07:50.400 --> 00:07:52.319 Rômulo, as the factory director, is 204 00:07:52.319 --> 00:07:55.280 funny to say factory director, right? 205 00:07:55.280 --> 00:07:57.120 A lot of oil, a lot... it gives the impression 206 00:07:57.120 --> 00:07:58.520 like we hear some metals 207 00:07:58.520 --> 00:08:02.440 clanging and such. I hear it, you hear it, but 208 00:08:02.440 --> 00:08:05.360 it's because I've seen us talk like that. 209 00:08:05.360 --> 00:08:07.080 So, you who are full of these crazy ideas 210 00:08:07.080 --> 00:08:10.000 let's not start there, there are 211 00:08:10.000 --> 00:08:12.039 You have to start here, you know? That's the 212 00:08:12.039 --> 00:08:14.919 initial process. So, there is a 213 00:08:14.919 --> 00:08:17.759 reading of someone else's business that we 214 00:08:17.759 --> 00:08:21.080 have obviously already learned how to do. 215 00:08:21.080 --> 00:08:24.120 Which, by the way, sets a starting point 216 00:08:24.120 --> 00:08:25.680 for us, right? Sometimes, people 217 00:08:25.680 --> 00:08:29.080 have a somewhat vague idea of 218 00:08:29.080 --> 00:08:32.279 what AI does, you know? No, it's not 219 00:08:32.279 --> 00:08:34.440 magic, on the contrary, right? We need to 220 00:08:34.440 --> 00:08:36.599 explain some obvious things to everyone and 221 00:08:36.599 --> 00:08:39.159 really understand these conditions of where 222 00:08:39.159 --> 00:08:40.599 we should start so that 223 00:08:40.599 --> 00:08:42.120 certain cognitive processes 224 00:08:42.120 --> 00:08:45.040 are already solved, so that later we 225 00:08:45.040 --> 00:08:47.279 can move on to the second stage of these 226 00:08:47.279 --> 00:08:49.480 projects, to the third stage of these 227 00:08:49.480 --> 00:08:52.519 projects. This is a quality, in a way, 228 00:08:52.519 --> 00:08:54.480 in the sense that we need to have 229 00:08:54.480 --> 00:08:56.920 this understanding of the business to 230 00:08:56.920 --> 00:08:59.880 be able to make the best assessment 231 00:08:59.880 --> 00:09:02.720 of where and for what reason we 232 00:09:02.720 --> 00:09:05.200 implement an FD first, right? Exactly. 233 00:09:05.200 --> 00:09:07.440 I think this idea of where fits with the 234 00:09:07.440 --> 00:09:10.320 question Lucca asked me, because many 235 00:09:10.320 --> 00:09:13.040 times we see AI there as 236 00:09:13.040 --> 00:09:16.120 just that chat on WhatsApp, right? 237 00:09:16.120 --> 00:09:18.640 All those AIs that Lucca mentioned from the 238 00:09:18.640 --> 00:09:20.519 Girra network, they're not on WhatsApp. 239 00:09:20.519 --> 00:09:22.560 They're inside the system, doing 240 00:09:22.560 --> 00:09:24.360 automated processes that no one is 241 00:09:24.360 --> 00:09:26.279 I see it happening in the vast majority of 242 00:09:26.279 --> 00:09:28.880 times, right? So when Lucca asks me 243 00:09:28.880 --> 00:09:31.560 about the Autobots, for example, about 244 00:09:31.560 --> 00:09:34.680 these backoffice AIs, this AI that stays 245 00:09:34.680 --> 00:09:37.839 running inside the systems, they are 246 00:09:37.839 --> 00:09:39.320 essential. We recently had 247 00:09:39.320 --> 00:09:41.240 recently in southern Brazil as 248 00:09:41.240 --> 00:09:43.760 a partner company of ours, right? A CRM that we 249 00:09:43.760 --> 00:09:45.640 are working on to automate 250 00:09:45.640 --> 00:09:47.640 practically all the processes within 251 00:09:47.640 --> 00:09:51.240 the CRM. So, the solution starts with the SDR 252 00:09:51.240 --> 00:09:53.399 which is the qualifier there. After 253 00:09:53.399 --> 00:09:56.200 it qualifies, all the stage movements 254 00:09:56.200 --> 00:09:58.360 of qualification. If this client is already 255 00:09:58.360 --> 00:10:00.800 with high probability of closing, low 256 00:10:00.800 --> 00:10:03.079 probability of closing, what are the 257 00:10:03.079 --> 00:10:04.880 products they want to buy, all the 258 00:10:04.880 --> 00:10:07.440 fields that the human salesperson would need 259 00:10:07.440 --> 00:10:08.839 to fill in, this is already filled 260 00:10:08.839 --> 00:10:11.200 automatically by the Autobots. Then 261 00:10:11.200 --> 00:10:14.800 it already schedules an appointment in the CRM calendar 262 00:10:14.800 --> 00:10:17.720 of the salesperson. And then the salesperson can still 263 00:10:17.720 --> 00:10:19.560 have a copilot. They click on a 264 00:10:19.560 --> 00:10:21.920 little button there, opens a button and they 265 00:10:21.920 --> 00:10:24.360 talk to the AI to resume a 266 00:10:24.360 --> 00:10:26.440 service, for example, with a client 267 00:10:26.440 --> 00:10:28.480 who was left with low qualification and 268 00:10:28.480 --> 00:10:31.079 sends one to automatic. There's a whole 269 00:10:31.079 --> 00:10:34.040 post-sales work they can do 270 00:10:34.040 --> 00:10:36.519 using AI. Anyway, the 271 00:10:36.519 --> 00:10:38.360 processes we've been creating with 272 00:10:38.360 --> 00:10:41.360 partner companies, they go far beyond 273 00:10:41.360 --> 00:10:43.720 WhatsApp, they go far beyond 274 00:10:43.720 --> 00:10:46.160 what's right in front of us. Of course, this 275 00:10:46.160 --> 00:10:48.480 stage of a qualifier, as you well 276 00:10:48.480 --> 00:10:51.200 said, Gustavo, it's usually one of the 277 00:10:51.200 --> 00:10:53.279 first steps we set up 278 00:10:53.279 --> 00:10:55.800 within the automation process, right 279 00:10:55.800 --> 00:10:57.519 when the digital employee is 280 00:10:57.519 --> 00:11:00.040 linked to sales, to reaching the public 281 00:11:00.040 --> 00:11:02.600 in general. Cool, Lucca, do you want to add 282 00:11:02.600 --> 00:11:05.399 something? In your speech, there is, uh, it 283 00:11:05.399 --> 00:11:07.639 caught my attention, sometimes we have to 284 00:11:07.639 --> 00:11:09.240 say no to the client, saying: Look, 285 00:11:09.240 --> 00:11:11.160 you want to start here, but this is not 286 00:11:11.160 --> 00:11:13.200 the first step. There's a big trap 287 00:11:13.200 --> 00:11:16.200 for engineers, you know, of 288 00:11:16.200 --> 00:11:18.880 smart ones, to automate a 289 00:11:18.880 --> 00:11:20.639 part, a process that didn't need to 290 00:11:20.639 --> 00:11:22.639 exist or that wasn't optimized 291 00:11:22.639 --> 00:11:24.560 so sometimes we rack our brains 292 00:11:24.560 --> 00:11:26.600 for a long time trying to create a 293 00:11:26.600 --> 00:11:28.200 automated solution, a solution that 294 00:11:28.200 --> 00:11:31.399 runs completely autonomously, 295 00:11:31.399 --> 00:11:33.720 when often that part of the 296 00:11:33.720 --> 00:11:36.519 workflow could be deleted because 297 00:11:36.519 --> 00:11:38.560 it's necessary with humans, but 298 00:11:38.560 --> 00:11:40.160 when you add technology to the 299 00:11:40.160 --> 00:11:41.760 process, sometimes it's not necessary 300 00:11:41.760 --> 00:11:43.920 or the way it's done today, it's not 301 00:11:43.920 --> 00:11:45.680 It's the most efficient way. So, why 302 00:11:45.680 --> 00:11:48.320 would I spend a lot of time putting it there 303 00:11:48.320 --> 00:11:50.920 technology that will repeat without making mistakes a 304 00:11:50.920 --> 00:11:53.560 process that is wrong from the start? So, 305 00:11:53.560 --> 00:11:55.639 this maturity to look at the processes, 306 00:11:55.639 --> 00:11:57.360 understand where we start, validate 307 00:11:57.360 --> 00:11:59.680 these processes so that later we just optimize and 308 00:11:59.680 --> 00:12:02.560 accelerate. This is something we bring 309 00:12:02.560 --> 00:12:05.200 to the table and, like, it's a diff... it's a 310 00:12:05.200 --> 00:12:06.560 very big differential. Consulting 311 00:12:06.560 --> 00:12:09.800 free. Well, ex, Lucca, just to wrap up, 312 00:12:09.800 --> 00:12:10.760 it costs 313 00:12:10.760 --> 00:12:14.440 expensive? Look, I think it's very cheap. 314 00:12:14.440 --> 00:12:16.480 Very cheap. It should cost much more 315 00:12:16.480 --> 00:12:18.760 expensive, but our mission is to put 316 00:12:18.760 --> 00:12:20.320 artificial intelligence in all 317 00:12:20.320 --> 00:12:22.600 companies in the world. We want a 318 00:12:22.600 --> 00:12:24.440 modest mission. It's a modest mission, a 319 00:12:24.440 --> 00:12:26.079 modest mission starting with a 320 00:12:26.079 --> 00:12:27.760 little country called Brazil with 200 million 321 00:12:27.760 --> 00:12:29.440 people. 322 00:12:29.440 --> 00:12:31.839 Look, for you to put an AI to 323 00:12:31.839 --> 00:12:34.399 work in your company tomorrow, starting 324 00:12:34.399 --> 00:12:38.360 from R$360 per month. That's it, folks, and then, 325 00:12:38.360 --> 00:12:40.880 let's suppose, I don't know, it's Christmas and you 326 00:12:40.880 --> 00:12:43.160 need to scale your customer service. 327 00:12:43.160 --> 00:12:46.360 You're running a campaign on Facebook, it 328 00:12:46.360 --> 00:12:47.880 scales naturally. You need 329 00:12:47.880 --> 00:12:50.639 more Santa Claus, for example, we clone 330 00:12:50.639 --> 00:12:52.839 from one moment to the next. Imagine if the 331 00:12:52.839 --> 00:12:55.000 If Santa Claus had a printer of 332 00:12:55.000 --> 00:12:58.000 elves that make the gifts, let's 333 00:12:58.000 --> 00:12:59.920 let's think about this logic 334 00:12:59.920 --> 00:13:01.600 Christmas logic here and 335 00:13:01.600 --> 00:13:04.360 yeah, no, I don't need it anymore, I'll download 336 00:13:04.360 --> 00:13:06.560 so, will I have to fire that 337 00:13:06.560 --> 00:13:07.920 person I trained? No, no, no, the 338 00:13:07.920 --> 00:13:09.079 digital employee can 339 00:13:09.079 --> 00:13:11.240 simply be reduced, it stays there 340 00:13:11.240 --> 00:13:13.240 almost hibernating and then eventually 341 00:13:13.240 --> 00:13:15.600 comes into play again when called, it's 342 00:13:15.600 --> 00:13:17.720 fantastic, really cool. Lucca, thanks 343 00:13:17.720 --> 00:13:19.959 for your explanations. Rômulo, director of the 344 00:13:19.959 --> 00:13:22.639 factory. Listen to the metals. 345 00:13:22.639 --> 00:13:25.279 Exactly. It was a pleasure having you here. 346 00:13:25.279 --> 00:13:27.839 To you as well, to you who listen to us, thank you very 347 00:13:27.839 --> 00:13:30.000 much. Follow us on all our 348 00:13:30.000 --> 00:13:31.440 social networks. We have been doing a 349 00:13:31.440 --> 00:13:34.720 really cool job, bringing information and 350 00:13:34.720 --> 00:13:37.160 education around the topic of 351 00:13:37.160 --> 00:13:38.920 artificial intelligence to all of 352 00:13:38.920 --> 00:13:43.000 Brazil. A hug, see you next time.