WEBVTT 1 00:00:00.400 --> 00:00:04.120 Welcome to part two. In part one, we uncovered in 2 00:00:04.120 --> 00:00:07.920 our interview with Oliver, founder of ADA AI, why the 3 00:00:07.920 --> 00:00:11.320 traditional back office is breaking under the weight of modern startup 4 00:00:11.320 --> 00:00:15.120 operations and how AI employees are already 5 00:00:15.200 --> 00:00:18.640 taking over full workflows at companies adopting 6 00:00:18.880 --> 00:00:22.480 agentic AI. But today we go even deeper 7 00:00:22.560 --> 00:00:26.320 in part two of this episode. Originally planners one episode, 8 00:00:26.320 --> 00:00:29.840 but Oliver was. Oliver was giving so good and long 9 00:00:29.840 --> 00:00:33.340 answers, we spontaneously decided to break it into two. 10 00:00:33.420 --> 00:00:36.820 In this episode, Dr. Oliver Lugos reveals the 11 00:00:36.820 --> 00:00:40.580 technical breakthroughs that separates simple chatbots from real 12 00:00:40.580 --> 00:00:44.140 autonomous agents. The industries that are secretly becoming 13 00:00:44.140 --> 00:00:47.859 agent ready. And why building an AI workforce might 14 00:00:47.859 --> 00:00:51.340 be the defining strategic advantage over the next decade. 15 00:00:51.980 --> 00:00:55.540 If you ever wondered how autonomous an AI should 16 00:00:55.540 --> 00:00:59.260 be, what God rays matter, or how far you 17 00:00:59.260 --> 00:01:03.060 can push automation without losing control, this is the episode 18 00:01:03.060 --> 00:01:06.200 you cannot miss. Let's jump into part two. 19 00:01:12.520 --> 00:01:15.560 Welcome to Startuprad IO, 20 00:01:16.120 --> 00:01:19.640 your podcast and YouTube blog covering the German 21 00:01:19.719 --> 00:01:23.160 startup scene with news, interviews and 22 00:01:23.400 --> 00:01:24.680 live events. 23 00:01:27.400 --> 00:01:31.120 Welcome to part two of our conversation with Dr. Oliver 24 00:01:31.120 --> 00:01:34.520 Lugosch, co founder of ADA AI and 25 00:01:34.520 --> 00:01:38.360 previously co founder of the Razer Group based in Berlin, the E commerce company 26 00:01:38.520 --> 00:01:42.120 that scaled beyond 700 million euros in annual 27 00:01:42.120 --> 00:01:45.960 revenue. In part one, Oliver walked us through the origins 28 00:01:45.960 --> 00:01:49.400 of ADACOP, the rise of AI employees and 29 00:01:49.480 --> 00:01:53.040 why Agentic AI is fundamentally different from 30 00:01:53.040 --> 00:01:56.440 every wave of automation that came before it. 31 00:01:56.600 --> 00:02:00.400 Today we shift gears. In the second segment, Oliver 32 00:02:00.400 --> 00:02:04.050 breaks down the technical and operational mechanics that 33 00:02:04.050 --> 00:02:07.850 allow AI employees to execute multi step 34 00:02:07.930 --> 00:02:11.530 workflows, make decisions, coordinate across tools and 35 00:02:11.770 --> 00:02:15.370 operate at levels that were considered impossible just a few years 36 00:02:15.370 --> 00:02:19.050 ago. We explored the real breakthroughs behind autonomous 37 00:02:19.050 --> 00:02:22.570 agents. How companies move from one agent to AI 38 00:02:22.570 --> 00:02:25.450 workforce, would guardrails prevent runaway autonomy, 39 00:02:25.850 --> 00:02:29.530 emerging industry leading the adoption curve, and 40 00:02:29.690 --> 00:02:33.130 how agentic AI will reshape high rank robos 41 00:02:33.430 --> 00:02:36.230 and organizational design. In part one 42 00:02:36.870 --> 00:02:40.710 showed us why AI employees matter and 43 00:02:40.710 --> 00:02:44.070 part two shows us what's coming next and how fast. 44 00:02:44.150 --> 00:02:47.750 Oliver, welcome back. Thank you. Great to be here. 45 00:02:48.390 --> 00:02:52.150 Totally my pleasure. For everybody listening, it was a few days or even 46 00:02:52.870 --> 00:02:56.550 longer for us, it has been just five minutes. So 47 00:02:57.590 --> 00:03:00.730 let us talk about what's the real technical 48 00:03:00.730 --> 00:03:03.450 breakthrough that makes AI employees 49 00:03:03.930 --> 00:03:07.290 fundamentally different from chatbots or those 50 00:03:07.290 --> 00:03:10.970 RPAs? Why we have chosen to 51 00:03:12.170 --> 00:03:15.970 call them AI employees is because that 52 00:03:15.970 --> 00:03:19.610 wording should show that 53 00:03:19.610 --> 00:03:23.210 the usage, the application, goes beyond these very 54 00:03:23.290 --> 00:03:26.900 singular purposes. What do I mean by that? A 55 00:03:26.900 --> 00:03:29.940 chatbot is there to chat, Right? It's mostly 56 00:03:30.580 --> 00:03:33.860 confined to the exchange of communication. Right? 57 00:03:34.900 --> 00:03:38.220 You can have small LLM helpers that for 58 00:03:38.220 --> 00:03:40.980 example, summarize documents. Right. And give you 59 00:03:41.940 --> 00:03:45.700 quick notes around it. But the thought about AI 60 00:03:45.700 --> 00:03:49.420 employees is that it can go much, much further, it can 61 00:03:49.420 --> 00:03:52.700 go beyond. It can actually act on a longer process 62 00:03:52.860 --> 00:03:56.140 scale, so to speak, and connect many different dots. 63 00:03:57.100 --> 00:04:00.780 And what you need for that is many different 64 00:04:00.780 --> 00:04:04.460 things. You need the capability of using LLMs in 65 00:04:04.460 --> 00:04:07.820 exactly the right way, that they produce the outcome that you want 66 00:04:07.980 --> 00:04:11.820 human like communication. You need integrations into many 67 00:04:11.820 --> 00:04:15.420 different systems. You need an integration into your communication system, 68 00:04:15.420 --> 00:04:18.620 into your Outlook or whatever email provider you use, 69 00:04:19.020 --> 00:04:22.850 into your maybe direct messaging service. You need integration 70 00:04:22.920 --> 00:04:26.520 integrations into your system of records, into your erp, your CRM, 71 00:04:26.680 --> 00:04:30.120 you know, for the logistics folks out there, into the tms, the wms, 72 00:04:31.080 --> 00:04:34.760 you need a human interface, you need a front 73 00:04:34.760 --> 00:04:38.440 end that the team can actually use, right? And 74 00:04:38.440 --> 00:04:42.160 there are all of these different elements that you need to combine in order 75 00:04:42.160 --> 00:04:45.840 to really think an entire process. Because only if you 76 00:04:45.840 --> 00:04:49.650 think an entire process, you can also have 77 00:04:49.650 --> 00:04:53.410 impact that is really sizable, that is really substantial. 78 00:04:53.810 --> 00:04:57.330 And it's more than a little helper here, a little 79 00:04:57.330 --> 00:05:00.930 helper there, where yes, it might be interesting, 80 00:05:01.090 --> 00:05:04.650 it might be helpful in everyday life, but it doesn't create 81 00:05:04.650 --> 00:05:07.890 this measurable meaningful impact for organization 82 00:05:08.210 --> 00:05:11.770 that truly, you know, gives teams the room to 83 00:05:11.770 --> 00:05:15.010 breathe and feel like, wow, something has really changed. 84 00:05:15.720 --> 00:05:19.480 And because you can combine all of these different things now, and 85 00:05:19.480 --> 00:05:23.120 because you can tie all of these ties together, you 86 00:05:23.120 --> 00:05:26.840 can think about a broader automation and more impact 87 00:05:27.320 --> 00:05:29.640 that creates benefit for the organization. 88 00:05:34.600 --> 00:05:37.800 As AI grows quickly, 89 00:05:38.120 --> 00:05:41.910 what patterns are you noticing in the demand 90 00:05:41.990 --> 00:05:45.830 of your customers? Which industries are or 91 00:05:46.070 --> 00:05:49.910 which processes are agent ready? Now, 92 00:05:50.070 --> 00:05:53.830 as of today, the pattern that. Emerges is 93 00:05:53.830 --> 00:05:57.630 that companies often have one or 94 00:05:57.630 --> 00:06:01.190 two use cases that is on top of everybody's minds because 95 00:06:01.590 --> 00:06:05.030 it seems to be present, it seems to be that people discuss these 96 00:06:06.150 --> 00:06:09.190 and we often start there. And they 97 00:06:10.060 --> 00:06:13.660 most frequently lie in the procurement process, 98 00:06:14.620 --> 00:06:18.220 sometimes in the distribution process, but often in procurement. 99 00:06:18.380 --> 00:06:21.900 And then when you go deeper and you start to work on the project and 100 00:06:21.900 --> 00:06:25.420 you start to build, then more and more people hear 101 00:06:25.500 --> 00:06:29.340 about, you know, what is possible and what can be done nowadays 102 00:06:29.340 --> 00:06:33.180 with the help of LLMs and other technology. And then 103 00:06:33.340 --> 00:06:37.100 more ideas emerge and people start to see more use cases 104 00:06:37.340 --> 00:06:40.700 from different parts of the organization and start to discuss 105 00:06:41.400 --> 00:06:44.960 and also initiate further project, further 106 00:06:44.960 --> 00:06:48.120 use cases across the org. So I would say 107 00:06:48.360 --> 00:06:52.120 classically your procure to pay and your order to cash 108 00:06:52.120 --> 00:06:55.600 flows are agent ready or automation 109 00:06:55.600 --> 00:06:59.160 ready? Really? I think the word AI 110 00:06:59.160 --> 00:07:02.920 agent is something that I find, I 111 00:07:02.920 --> 00:07:06.440 don't know, difficult in how to interpret it. Right. 112 00:07:06.760 --> 00:07:10.600 What does an agent Actually do. What does an agent mean? I feel 113 00:07:10.600 --> 00:07:14.300 like it's a word that everybody uses. But what is really behind 114 00:07:14.300 --> 00:07:17.380 that, behind that curtain? What, what, what is an agent? 115 00:07:18.420 --> 00:07:22.260 I would bet if you ask 10 people, you get 12 answers. 116 00:07:25.860 --> 00:07:29.700 That always reminds me, in the, in the mid-1990s, 117 00:07:29.700 --> 00:07:33.060 everybody was trying to sell multi multimedia 118 00:07:33.300 --> 00:07:36.820 computers and they had experts from like five new 119 00:07:36.900 --> 00:07:40.420 newspapers doing the elevator pitch and everybody talked about something 120 00:07:40.420 --> 00:07:44.240 different. Yeah, I fully, fully agree. Right. And 121 00:07:44.240 --> 00:07:47.960 by the way, I'm not claiming that AI employee is the best wording, right. In 122 00:07:47.960 --> 00:07:51.800 any way. I'm not claiming that. I'm just saying that, you know, these AI agents 123 00:07:51.800 --> 00:07:55.280 have, have popped up and I find it difficult to grasp what is really, 124 00:07:55.760 --> 00:07:59.520 really meant by it. In, in any case, I think 125 00:08:00.640 --> 00:08:04.440 these use cases pop up throughout the classical again, procure 126 00:08:04.440 --> 00:08:08.240 to pay, order to cash cycles also on the money flow, 127 00:08:08.240 --> 00:08:11.440 right. Speaking about finance, order invoice processing, 128 00:08:12.230 --> 00:08:15.670 issuing of invoices, reconciliation accounting, also these 129 00:08:15.670 --> 00:08:19.470 processes are ready. What you shouldn't forget 130 00:08:19.470 --> 00:08:23.270 is that this human in the loop process is something that 131 00:08:23.270 --> 00:08:26.790 you typically start from, right? If you want to automate a process, 132 00:08:27.110 --> 00:08:30.950 do it, but still have a human confirm certain actions, 133 00:08:30.950 --> 00:08:34.630 confirm certain communication that goes out. So you can already 134 00:08:36.070 --> 00:08:39.790 get 80% of the efficiency of 135 00:08:39.790 --> 00:08:43.580 the time saving, maybe even 90%. But you still have that 136 00:08:43.810 --> 00:08:47.250 component of a human double checking and ultimately signing off 137 00:08:47.410 --> 00:08:51.210 on certain actions. And that is typically the starting point. And then when you 138 00:08:51.210 --> 00:08:54.850 want to move to full automation, get these last 20 to 139 00:08:54.850 --> 00:08:58.370 10% of, of, you know, 140 00:08:58.370 --> 00:09:01.810 capacity of efficiency. That's what you should do under certain 141 00:09:01.810 --> 00:09:05.530 circumstances, right? If you really have the trust, if you really have seen 142 00:09:05.530 --> 00:09:09.090 a lot of positive examples, a lot of perfect 143 00:09:09.790 --> 00:09:13.390 drafts and proposals from the AI, that's when you can move to 144 00:09:13.390 --> 00:09:16.910 full automation. But that's a very, very organic process. 145 00:09:16.910 --> 00:09:20.630 That's a very trust driven process that just comes over 146 00:09:20.630 --> 00:09:24.190 time and again. Even if you are stuck with these, you know, 147 00:09:24.510 --> 00:09:28.270 human in the loop processes, you still have most of the value 148 00:09:28.270 --> 00:09:30.110 that will, that you will get ultimately. 149 00:09:36.050 --> 00:09:39.610 We've been always talking about the employees who get rid of very boring 150 00:09:39.610 --> 00:09:43.290 stuff. We've been talking about the entrepreneurs who get more 151 00:09:43.290 --> 00:09:46.610 efficient with a few employees. But where are the 152 00:09:46.610 --> 00:09:50.290 customer wins? What's the most surprising outcome a 153 00:09:50.290 --> 00:09:54.050 customer has seen after deploying an AI 154 00:09:54.130 --> 00:09:57.330 employee? The first thing that customers think about 155 00:09:57.810 --> 00:10:01.650 is the efficiency, right? And saving capacity, 156 00:10:01.730 --> 00:10:05.390 saving resources. But what also comes out 157 00:10:05.390 --> 00:10:08.390 of, of all of these projects is 158 00:10:09.830 --> 00:10:13.390 an improved speed of processing, data processing 159 00:10:13.390 --> 00:10:16.710 input, responding to customers, responding to suppliers, 160 00:10:17.030 --> 00:10:20.830 reacting to any kind of signal that comes in really. So 161 00:10:20.830 --> 00:10:24.590 you get rid of this backlog, right, that often piles up and it 162 00:10:24.590 --> 00:10:28.190 can again slow down other parts of the company. Other 163 00:10:28.190 --> 00:10:31.100 teams will notice, will notice that 164 00:10:31.420 --> 00:10:34.780 processes run quicker and more reliably 165 00:10:35.260 --> 00:10:38.540 with the use of these automations. And the 166 00:10:38.540 --> 00:10:42.300 quality of the output data is typically much better. 167 00:10:42.620 --> 00:10:46.380 If anything, it's certainly of higher consistency, 168 00:10:46.540 --> 00:10:49.580 right? Because if you use an AI based automation, 169 00:10:49.820 --> 00:10:53.580 there's consistency of how things are interpreted, of how things 170 00:10:53.580 --> 00:10:57.300 are put into a certain place in the system, for example, and 171 00:10:57.300 --> 00:11:00.770 that consistency drives usability of that data 172 00:11:00.850 --> 00:11:04.450 because, you know, it's no longer, I would say a human spread 173 00:11:04.450 --> 00:11:08.090 of how things are being done, but it's typically very clean and 174 00:11:08.090 --> 00:11:10.530 very standardized. And that is 175 00:11:11.890 --> 00:11:15.730 a benefit that many companies don't foresee, but 176 00:11:15.730 --> 00:11:19.410 that they experience as they use the AI employees because 177 00:11:19.410 --> 00:11:23.170 they make the data more consistent and more usable going 178 00:11:23.170 --> 00:11:24.380 forward. Also for other teams. 179 00:11:28.050 --> 00:11:31.650 I see you do that because you want to scale. 180 00:11:31.970 --> 00:11:35.450 But what are some scaling challenges with 181 00:11:35.450 --> 00:11:37.570 AI employees? What's the most 182 00:11:37.970 --> 00:11:41.650 underestimated challenge that companies, your 183 00:11:41.650 --> 00:11:45.250 customers face when scaling from one agent 184 00:11:45.250 --> 00:11:48.530 to an AI workforce, to many agents? 185 00:11:49.250 --> 00:11:52.970 When we think about these automations and changing 186 00:11:52.970 --> 00:11:56.280 processes in companies, ultimately that's what we're doing, we're changing processes. 187 00:11:56.350 --> 00:11:59.870 Processes. Then you should never underestimate the human 188 00:11:59.870 --> 00:12:03.350 factor, right? You will always have those teams that are 189 00:12:03.350 --> 00:12:06.910 responsible for the processes, that oversee the processes, that 190 00:12:06.910 --> 00:12:10.590 further develop and advance the processes. And 191 00:12:10.990 --> 00:12:14.350 you cannot go faster than the team can go, 192 00:12:14.510 --> 00:12:18.350 right? You always have to take everybody along and make sure 193 00:12:18.430 --> 00:12:22.110 that these changes happen at a pace that is digestible 194 00:12:22.110 --> 00:12:25.610 for the organization. And so the limitation 195 00:12:26.170 --> 00:12:29.690 in scaling is really not a technical limitation. 196 00:12:30.010 --> 00:12:33.610 It is not a limitation of, you know, being able to produce these 197 00:12:33.610 --> 00:12:37.250 automations. It's a limitation of how quickly can 198 00:12:37.250 --> 00:12:41.090 teams and organizations adopt to these changes. And I think that's a very, 199 00:12:41.090 --> 00:12:44.890 very natural limitation. And it's something that you 200 00:12:44.890 --> 00:12:47.970 should always keep in mind when you think about how quickly do I want to 201 00:12:47.970 --> 00:12:51.800 scale from 1 to 2 to 5 to 10 use 202 00:12:51.800 --> 00:12:53.360 cases or AI employees. 203 00:12:55.760 --> 00:12:59.280 I see in the founders world we will be talking about 204 00:12:59.440 --> 00:13:02.880 what's the workflow. No founder realizes can be 205 00:13:02.880 --> 00:13:06.640 automated, but every founder should automate using 206 00:13:06.880 --> 00:13:10.640 AI employee. Let's do a little bit 207 00:13:10.640 --> 00:13:14.480 outlook into the future. How do you see the balance 208 00:13:14.480 --> 00:13:17.810 between human teams and AI employees 209 00:13:18.370 --> 00:13:21.890 evolving over let's say the next five years? I know 210 00:13:22.210 --> 00:13:25.810 it's pretty, pretty early, but givens a positive vision, 211 00:13:27.410 --> 00:13:30.370 I think this. View is pretty clear. 212 00:13:31.170 --> 00:13:34.610 I think there will be less and less and less of these 213 00:13:35.250 --> 00:13:38.930 painful Manual things that you have to do, the repetitive ones. 214 00:13:39.010 --> 00:13:42.610 I think we will spend more time on really figuring out stuff, 215 00:13:42.690 --> 00:13:46.460 figuring out how to do things and then having your AI 216 00:13:46.460 --> 00:13:50.060 employee, your assistant, your agent, whatever you want to call it, take 217 00:13:50.060 --> 00:13:53.780 over what you have defined. It 218 00:13:54.100 --> 00:13:57.620 will also help us become more creative. 219 00:13:57.860 --> 00:14:01.140 You can use LLMs to get new ideas, get new 220 00:14:01.140 --> 00:14:04.580 impulses and then turn them into great ideas, turn them into great 221 00:14:04.580 --> 00:14:08.340 execution and really move your business, 222 00:14:08.420 --> 00:14:12.200 your startup, to the next level. So I think you will 223 00:14:12.200 --> 00:14:15.760 have less of this boring, painful piece and you will have 224 00:14:15.760 --> 00:14:19.320 more of the creativity. Do things differently 225 00:14:19.720 --> 00:14:23.320 and, you know, come into a state where it's, it's, it's 226 00:14:23.320 --> 00:14:27.080 exciting to advance your business because you no longer have to take 227 00:14:27.080 --> 00:14:29.080 care of the very, very basic stuff. 228 00:14:31.560 --> 00:14:35.240 I was just wondering about AI employees. The 229 00:14:35.480 --> 00:14:39.240 very, very boring stuff. I was wondering, would it be a dream 230 00:14:39.240 --> 00:14:42.890 for you to automate or start 231 00:14:42.970 --> 00:14:46.650 automation? German tax authorities but 232 00:14:46.650 --> 00:14:49.930 because I do believe there will be quite a lot of potential 233 00:14:50.330 --> 00:14:54.010 and I also do believe there are a lot of people who 234 00:14:54.170 --> 00:14:58.010 have jobs that are very repetitive and I'm sure 235 00:14:58.090 --> 00:15:01.810 they would love to get rid of at least those steps 236 00:15:01.810 --> 00:15:05.290 of the work. I'm not a 237 00:15:05.370 --> 00:15:08.730 tax expert, but what I can say as, you know, 238 00:15:09.260 --> 00:15:13.100 being subject to tax, I fully agree with you. I think 239 00:15:13.260 --> 00:15:16.980 there are probably many, many process steps in, you know, 240 00:15:16.980 --> 00:15:20.380 filing taxes, anything around taxes that 241 00:15:20.620 --> 00:15:23.660 could be at least simplified or 242 00:15:23.660 --> 00:15:27.420 accelerated with AI and LLMs. And 243 00:15:27.500 --> 00:15:31.260 I think there is a huge space that is to be tackled in that area. 244 00:15:31.260 --> 00:15:34.460 I fully agree now around the regulation 245 00:15:35.510 --> 00:15:39.190 aspect of it. I cannot speak to it, but I'm sure 246 00:15:39.190 --> 00:15:42.310 we'll figure this out and make sure that, you know, 247 00:15:43.910 --> 00:15:47.470 we can also use AI in these, in these topics 248 00:15:47.470 --> 00:15:49.510 and these areas of businesses agree. 249 00:15:52.310 --> 00:15:55.910 Going back into the, from government into the 250 00:15:55.910 --> 00:15:59.590 real industries. We've been talking about 251 00:16:01.120 --> 00:16:04.880 AI agents helping employees to 252 00:16:04.880 --> 00:16:08.240 get rid of boring work. But I was 253 00:16:08.240 --> 00:16:11.680 wondering, how do you imagine 254 00:16:12.080 --> 00:16:15.760 agentic AI reshaping the org 255 00:16:15.760 --> 00:16:19.520 charts of larger companies? I think 256 00:16:19.520 --> 00:16:23.200 AI will start to become part of the org charts. 257 00:16:23.440 --> 00:16:26.640 Right. Ultimately, that's where we're headed. 258 00:16:27.180 --> 00:16:30.860 And what does it resemble? Org charts basically resemble 259 00:16:31.420 --> 00:16:34.620 or mirror, you know, certain resources 260 00:16:34.700 --> 00:16:38.460 capacities. I'm sorry, I'm not trying to be, you know, inhumane or anything. 261 00:16:38.620 --> 00:16:42.260 I'm just trying to abstract it. Right. What is mirrored in an 262 00:16:42.260 --> 00:16:45.260 org chart, it is, you know, human resources 263 00:16:45.740 --> 00:16:48.940 executing certain processes, certain tasks, taking over certain 264 00:16:48.940 --> 00:16:52.580 responsibilities. As we see AI applications take 265 00:16:52.580 --> 00:16:56.360 over that role, they move into that capacity. Right. And 266 00:16:56.360 --> 00:16:59.800 you will see that a person will be completely 267 00:16:59.800 --> 00:17:02.880 occupied with running many, many, many 268 00:17:03.040 --> 00:17:06.880 operational or a large volume of operations through 269 00:17:06.880 --> 00:17:10.640 the help of AI, AI employees, agents, whatever you want to call them, 270 00:17:10.960 --> 00:17:14.720 and being that person that keeps an eye on all of 271 00:17:14.720 --> 00:17:18.200 these processes, right, keeps an eye on quality, further 272 00:17:18.200 --> 00:17:21.620 advances the process, further advances the model, the 273 00:17:21.620 --> 00:17:25.100 application. And really being able to 274 00:17:25.420 --> 00:17:29.100 have such a huge leverage, right? One person, imagine one 275 00:17:29.100 --> 00:17:32.780 person handling, I don't know, thousands of customer requests every day. 276 00:17:32.860 --> 00:17:36.620 Why? Because they have LLMs and AI at their fingertips, 277 00:17:36.620 --> 00:17:40.180 right? And they can do this at a speed and a 278 00:17:40.180 --> 00:17:43.820 quality that was unprecedented, that you could never have had with 279 00:17:44.140 --> 00:17:47.660 just humans. Because you can further improve 280 00:17:47.900 --> 00:17:51.710 these algorithms and these automations more and more and more. And 281 00:17:51.710 --> 00:17:55.470 I think that operational leverage that you can have will result in org 282 00:17:55.470 --> 00:17:58.950 charts looking very different from today. Do you know 283 00:17:59.030 --> 00:18:02.750 when you've been talking about the difference in org 284 00:18:02.750 --> 00:18:06.590 charts, what do you say triggered one idea in 285 00:18:06.590 --> 00:18:10.390 my mind because when I was working in a lot 286 00:18:10.390 --> 00:18:14.230 of different companies as a consultant, if you're a good consultant, you get 287 00:18:14.230 --> 00:18:17.910 invited to the Christmas party of the company and 288 00:18:17.910 --> 00:18:21.710 then you do have the retirees, the people who come in, they 289 00:18:21.710 --> 00:18:25.243 retired and they talk about what was work like 5, 10, 290 00:18:25.310 --> 00:18:28.610 20 years ago. And I was wondering if at one point 291 00:18:29.010 --> 00:18:32.730 you'll have like one room in a company. I 292 00:18:32.730 --> 00:18:35.970 had in mind a lot of old screens where you can basically talk to old 293 00:18:35.970 --> 00:18:39.570 AI agents and, and pick their brains 294 00:18:39.650 --> 00:18:43.410 on knowledge from the past or how it was done in the past. 295 00:18:43.730 --> 00:18:47.530 Why not? I think it would be enriching, right, if you could tap into 296 00:18:47.530 --> 00:18:51.380 that knowledge. It's still fresh because it will always be fresh, right? It 297 00:18:51.380 --> 00:18:55.060 will remember and we'll be able to help us and learn, right. 298 00:18:55.300 --> 00:18:58.860 Obviously you will never be able to fully take over 299 00:18:58.860 --> 00:19:02.460 that human element, right? You will always have the core 300 00:19:02.460 --> 00:19:06.060 team, will be humans at your Christmas party and you will probably 301 00:19:06.060 --> 00:19:09.620 not celebrate with AI employees. I don't know, maybe we do, 302 00:19:09.780 --> 00:19:13.140 I assume we will not. But having that, 303 00:19:13.860 --> 00:19:17.520 you know, this, these capacities of remembering stuff very 304 00:19:17.520 --> 00:19:20.960 clearly giving you insights, you can ask any questions it will 305 00:19:20.960 --> 00:19:24.640 hopefully answer in the correct way, in a true way. 306 00:19:25.360 --> 00:19:28.920 I think this is a value add that we are just starting 307 00:19:28.920 --> 00:19:31.360 to understand, just starting to grasp really 308 00:19:32.560 --> 00:19:35.840 what I. Had in mind when I was talking about this was having a digital 309 00:19:35.920 --> 00:19:39.680 room, not necessarily to be entered during Christmas party. We all know that 310 00:19:39.680 --> 00:19:43.120 can get out of hands. A digital room where can 311 00:19:43.120 --> 00:19:46.620 basically look into different screens and have 312 00:19:46.620 --> 00:19:50.380 all the old AI agent archived there, which 313 00:19:50.540 --> 00:19:54.260 of course will be a big job for knowledge management. I'm sure you're working on 314 00:19:54.260 --> 00:19:55.580 IDs for that as well. 315 00:19:57.660 --> 00:20:01.100 Let's pick your brain a little bit about a contrarian view. 316 00:20:01.580 --> 00:20:05.300 What is one unpopular truth about autonomous 317 00:20:05.300 --> 00:20:08.900 agents that nobody in the AI industry wants to 318 00:20:08.900 --> 00:20:12.540 admit? I think it comes down to this 319 00:20:12.620 --> 00:20:16.340 topic of autonomy that you mentioned before. I 320 00:20:16.340 --> 00:20:19.880 think the unpopular truth is that we are 321 00:20:19.880 --> 00:20:23.440 still far away from this true, true fundamental 322 00:20:23.440 --> 00:20:26.360 autonomy of really running human processes. 323 00:20:27.480 --> 00:20:31.160 People sometimes claim and continue to claim that we are there 324 00:20:31.160 --> 00:20:34.360 already. I don't see it. I have not seen it in 325 00:20:34.520 --> 00:20:38.240 application and I think it's still a little bit of 326 00:20:38.240 --> 00:20:41.960 a way to go to really have that live 327 00:20:42.120 --> 00:20:44.040 and ready and in production. 328 00:20:47.350 --> 00:20:50.790 For our founders out there, for our audience. DM us on 329 00:20:50.790 --> 00:20:53.510 LinkedIn, me, your manager. 330 00:20:54.710 --> 00:20:57.950 What would be the one process you'd 331 00:20:57.950 --> 00:21:01.630 automate first? We'll share the most creative ones 332 00:21:01.630 --> 00:21:02.710 anonymously. 333 00:21:05.190 --> 00:21:08.950 Looking forward to ideas there. Let's get a 334 00:21:08.950 --> 00:21:12.640 little bit towards the end and this pencil a little bit at end. The 335 00:21:12.790 --> 00:21:16.070 advice, what's 336 00:21:16.470 --> 00:21:19.590 your advice to early stage founders who want to build 337 00:21:20.550 --> 00:21:24.190 AI ready operational structures when they're 338 00:21:24.190 --> 00:21:26.630 starting out like today or. 339 00:21:27.670 --> 00:21:31.350 Well, Today is late November 2025, but this will 340 00:21:31.350 --> 00:21:35.150 air in January 2026. A lot of activity is going on in 341 00:21:35.150 --> 00:21:38.950 January. Everybody wants to start something new. What advice would you give 342 00:21:38.950 --> 00:21:42.640 to entrepreneurs now thinking about setting up the company? 343 00:21:44.400 --> 00:21:48.160 Great question. I think yes, it should be 344 00:21:48.400 --> 00:21:52.040 AI native, I think because not starting something 345 00:21:52.040 --> 00:21:55.280 AI native or with AI in mind would be 346 00:21:55.520 --> 00:21:58.920 foolish given all of the technology that we have and the 347 00:21:58.920 --> 00:22:02.520 capabilities the technology brings. But I won't, I wouldn't 348 00:22:02.520 --> 00:22:06.160 overdo it. I wouldn't say, you know, first day you start 349 00:22:06.160 --> 00:22:09.560 thinking about AI. No, first day think what is your business? What do you want 350 00:22:09.560 --> 00:22:13.300 to do? What's the value that you will bring to your 351 00:22:13.300 --> 00:22:17.100 customers, to, you know, society, to the world. And since post 352 00:22:17.100 --> 00:22:20.580 corona, how do you make money? What's your unit? 353 00:22:20.580 --> 00:22:24.420 Economics, your mix of scale. And then think about 354 00:22:24.500 --> 00:22:28.300 how I could raise that. That would 355 00:22:28.300 --> 00:22:31.860 be something I will be thinking about. How about you, Oliver? Exactly 100%. 356 00:22:31.940 --> 00:22:34.740 So first start with, you know, what are you doing? How do you create app 357 00:22:34.740 --> 00:22:38.580 value? How do you monetize? All of these are very, very important questions. And 358 00:22:38.580 --> 00:22:42.020 once you figure that out, or once you have an hypothesis and you want to 359 00:22:42.020 --> 00:22:45.830 test it, use AI as a tool, right? Use LLMs to then 360 00:22:45.830 --> 00:22:49.190 scale yourself, scale your time. But 361 00:22:50.150 --> 00:22:53.590 if you're not a business, you know that, that, 362 00:22:54.390 --> 00:22:57.670 that sells AI where AI is really the core product. 363 00:22:58.469 --> 00:23:02.110 Don't start with AI the very first day. Start with the fundamentals of the 364 00:23:02.110 --> 00:23:05.950 business. I think this is true. This will always be true. And once you've started 365 00:23:05.950 --> 00:23:09.510 to figure out things and you have your hypotheses or something that's even 366 00:23:09.510 --> 00:23:12.950 proven, use AI to scale. Use AI to 367 00:23:13.030 --> 00:23:16.470 create that leverage that you want to have. 368 00:23:20.150 --> 00:23:23.950 I'm curious for one thing, because we go 369 00:23:23.950 --> 00:23:27.350 into the last standard question, and then we have the two mandatory. But 370 00:23:27.510 --> 00:23:31.030 before we go into that, can you see a future 371 00:23:31.350 --> 00:23:35.030 where the AI employees are targeted 372 00:23:35.030 --> 00:23:38.630 by hackers and used for nefarious purposes? 373 00:23:40.220 --> 00:23:43.740 I think that time is already here. I'm very sure that 374 00:23:43.740 --> 00:23:47.380 hackers are already trying to attack LLM systems, 375 00:23:47.380 --> 00:23:51.060 AI systems. And it's something that you have to stay on 376 00:23:51.060 --> 00:23:54.580 top of and that you have to build 377 00:23:54.580 --> 00:23:58.380 countermeasures against. So I think that's absolutely critical. 378 00:23:58.380 --> 00:24:02.020 Whenever you design any kind of AI system that you have that in mind, 379 00:24:02.020 --> 00:24:05.660 people will always try to get to your system. They will 380 00:24:05.660 --> 00:24:09.460 always try to hack you. So security is 381 00:24:09.540 --> 00:24:13.340 number one priority, particularly if you think in broad applications. 382 00:24:13.340 --> 00:24:17.100 Right. For a small, small company, small startup, they might not 383 00:24:17.100 --> 00:24:20.620 be the first target, but when you move to larger scales, then you're 384 00:24:20.620 --> 00:24:24.299 certainly targeted day in, day out. And that must be very core 385 00:24:24.299 --> 00:24:28.100 to how you approach the entire thing. I 386 00:24:28.100 --> 00:24:31.700 see, I see. And the 387 00:24:31.700 --> 00:24:35.220 last question I prepared for this interview, if every company 388 00:24:35.460 --> 00:24:39.050 in the world would deploy AI employees tomorrow, 389 00:24:39.610 --> 00:24:42.010 what do you guess would be the one that breaks first? 390 00:24:44.330 --> 00:24:48.010 Like what function? What capability? 391 00:24:48.090 --> 00:24:51.770 What would break first? Great question, 392 00:24:52.010 --> 00:24:55.650 great question. I mean, looking at what LLMs are great 393 00:24:55.650 --> 00:24:59.250 at, I would assume some of the numerical stuff would 394 00:24:59.250 --> 00:25:02.890 probably break. I don't know, maybe the forecasting piece or 395 00:25:03.210 --> 00:25:06.800 the optimization piece might be something that is 396 00:25:06.800 --> 00:25:10.480 most fragile because, you know, LLMs, 397 00:25:10.560 --> 00:25:13.440 large language models, are not designed for 398 00:25:14.000 --> 00:25:17.600 numerical analytical approaches. But that's just my guess. 399 00:25:18.560 --> 00:25:21.720 I hope we will not see that happen. But if I had to guess that, 400 00:25:21.720 --> 00:25:22.640 that would be my take. 401 00:25:25.760 --> 00:25:29.480 Our standard questions are pretty, pretty normal. Are you open to talk 402 00:25:29.480 --> 00:25:33.010 to new investors? We are always open to talk to new 403 00:25:33.010 --> 00:25:36.810 investors. We are not actively fundraising, but we are always 404 00:25:36.810 --> 00:25:39.290 open to talk. And 405 00:25:40.810 --> 00:25:44.650 this is a funny question for you, because are you also open 406 00:25:44.650 --> 00:25:47.610 to look for talented employees that you cannot yet, 407 00:25:49.370 --> 00:25:53.050 that you cannot completely replace with AI employees? 408 00:25:53.610 --> 00:25:57.410 We are also looking for great people. Absolutely. At any point in 409 00:25:57.410 --> 00:26:01.130 time. So, yes, we are hiring. So 410 00:26:01.130 --> 00:26:04.810 basically, we'll link down here your career website and your personal 411 00:26:04.810 --> 00:26:08.610 LinkedIn profile so investors can reach out to you. Perfect. Let's do that. 412 00:26:09.330 --> 00:26:13.090 Oliver, thank you very much. Was such a pleasure to have you here twice. 413 00:26:13.170 --> 00:26:16.890 Thank you. Thank you, Johan. It was a real pleasure. Was great talking to 414 00:26:16.890 --> 00:26:20.050 you. Same here. Have a good day. Bye bye. Thanks. 415 00:26:25.570 --> 00:26:29.180 That's all folks. Find more news streams, 416 00:26:29.340 --> 00:26:30.500 events and 417 00:26:30.500 --> 00:26:34.780 interviews@www.startuprad.IO. 418 00:26:35.180 --> 00:26:37.180 remember, sharing is car.