WEBVTT 00:00:00.000 --> 00:00:07.920 Music. 00:00:08.922 --> 00:00:14.842 Your podcast and YouTube blog covering the German startup scene with news, 00:00:15.082 --> 00:00:17.742 interviews, and live events. 00:00:19.982 --> 00:00:23.802 Hello and welcome, everybody. This is Joe from StartupRate.io, 00:00:23.982 --> 00:00:26.742 your startup podcast and YouTube blog from Germany. 00:00:26.982 --> 00:00:33.222 Today, I welcome another guest in our interview series sponsored by Hassan Trade & Invest. 00:00:33.402 --> 00:00:37.102 I would like to welcome Patrick here from Frankfurt. Hey, how are you doing? 00:00:37.762 --> 00:00:41.742 Hey, Joel. Thanks. I'm feeling great, and I'm very happy to be here. 00:00:42.422 --> 00:00:47.302 It's good that you feel great, because we had to move a little bit our recording 00:00:47.302 --> 00:00:51.382 due to me being sick, you being sick, and now finally we made it, 00:00:51.442 --> 00:00:53.262 I think, after eight weeks? 00:00:54.002 --> 00:00:57.302 Something like that, yes. Something like that. Yes, great. 00:00:57.742 --> 00:01:02.462 Let's talk a little bit about our Enabler HTAI and the Enterprise Europe Network 00:01:02.462 --> 00:01:07.642 Hessen. This recording was made possible by HTII and the Enterprise Europe Network Hessen. 00:01:08.102 --> 00:01:12.842 These organizations have made tremendous contributions to helping startup businesses 00:01:12.842 --> 00:01:18.322 succeed and thrive, providing a range of services from helping to find grants 00:01:18.322 --> 00:01:19.682 to ongoing partnerships. 00:01:19.962 --> 00:01:24.902 By taking advantage of these resources, startup companies can network and develop 00:01:24.902 --> 00:01:28.822 innovative strategies for success on the international stage. 00:01:28.822 --> 00:01:35.122 The dedicated support of HTIA and the EEN Hessen is paramount in providing startup 00:01:35.122 --> 00:01:39.402 businesses with the tools for lasting success. You can learn more in the links 00:01:39.402 --> 00:01:41.682 down here in the show notes. 00:01:42.282 --> 00:01:47.762 So, Patrick, we met in the past, we have realized. 00:01:48.102 --> 00:01:53.822 And if we're talking about the past, I mean like ancient times in terms of the startup world, right? 00:01:54.502 --> 00:02:01.802 Absolutely. If I recall correctly, we met at the Startup Week in Frankfurt that 00:02:01.802 --> 00:02:04.862 must have been over 10 years ago, something like that. 00:02:05.162 --> 00:02:10.522 I think it was a Startup Weekend like nine years ago, but no big difference at all. 00:02:14.094 --> 00:02:21.234 Fun fact that I found my first co-founder from a very first startup at this very startup week. 00:02:21.454 --> 00:02:25.674 That was actually like really my kickoff into the startup world and being a founder. 00:02:25.954 --> 00:02:30.654 A big mental high five to Mario Hachma, one of our co-hosts who actually organized 00:02:30.654 --> 00:02:33.314 this startup weekend. Thank you very much. 00:02:33.874 --> 00:02:37.174 Big shout out to Mario in any case. He's awesome. Yes. 00:02:38.014 --> 00:02:42.234 Today we are here to talk a little bit about who you are. 00:02:42.354 --> 00:02:46.474 What you did, and OmniFact, an AI startup. 00:02:47.614 --> 00:02:54.474 But first, I was going a little bit through your CV, and I see you are actually 00:02:54.474 --> 00:02:56.774 a developer at heart. Is that true? 00:02:57.234 --> 00:03:02.474 That is true. That is true. I studied computer science and business, 00:03:02.754 --> 00:03:04.294 but I only finished computer science. 00:03:04.454 --> 00:03:09.234 You've done a little bit of financial analyst, working student, student worker. 00:03:09.894 --> 00:03:16.494 But then, from what I can see, you've been an entrepreneur all your professional 00:03:16.494 --> 00:03:19.834 life, going back to 2012. Is that true? 00:03:20.214 --> 00:03:26.234 That is true. I studied law for like one semester. Then I saw that that wasn't for me. 00:03:26.894 --> 00:03:33.134 Then I did what I loved. I became a programmer. Then I thought I had to study something. 00:03:33.614 --> 00:03:38.794 Then I started business, which kind of bored me. So I went to computer science. 00:03:39.234 --> 00:03:46.114 And during my studies, I started a small boutique for pen testing, 00:03:46.174 --> 00:03:48.194 actually, and social engineering. 00:03:48.514 --> 00:03:52.974 It actually came from the hacker scene and the security scene in the very beginning. 00:03:53.881 --> 00:03:57.141 What did you do there, and are you currently social engineering us? 00:03:58.601 --> 00:04:02.041 If I would tell you, where would be the fun? 00:04:04.241 --> 00:04:10.961 What we did there was basically we tried to help mid-sized German companies 00:04:10.961 --> 00:04:12.281 to secure their network. 00:04:12.281 --> 00:04:19.121 And we especially helped executives who traveled to countries like Israel, China, 00:04:19.881 --> 00:04:27.761 Japan, Middle East, where you have actual risks of being targeted by state actors 00:04:27.761 --> 00:04:33.461 or large organizations to be more secure in their travels and their daily lives. 00:04:34.741 --> 00:04:36.361 And we were really lucky. We 00:04:36.361 --> 00:04:41.341 started, I think, like four weeks in or eight weeks in, Snowden came out. 00:04:42.281 --> 00:04:49.201 And that basically was the best push for any starting business you could imagine. I see. 00:04:49.721 --> 00:04:55.401 And then at one point, you started building AI and software solutions for other people. 00:04:56.321 --> 00:05:02.361 Yes. So from there, after the startup weekend, I made Khan, 00:05:03.121 --> 00:05:08.241 which still is a great friend of mine, with whom I started SecDash, 00:05:08.361 --> 00:05:11.481 which was like a little bit between today and the past. 00:05:11.481 --> 00:05:18.921 It was a startup to monitor the chatter of IT security issues in the dark web 00:05:18.921 --> 00:05:24.481 and IRC channels and so on to basically get an early warning system for new 00:05:24.481 --> 00:05:27.641 attacks on websites, on servers and so on. 00:05:28.061 --> 00:05:32.881 And that was actually the point where I for the first time worked with natural language processing. 00:05:33.121 --> 00:05:37.141 So that basically was the first time was working with the technologies that 00:05:37.141 --> 00:05:40.861 later became what we currently call AI. 00:05:42.561 --> 00:05:48.581 And that startup failed miserably on the business side, but it was a great learning experience. 00:05:49.441 --> 00:05:53.901 And during that time, I met Florian and Alex. 00:05:54.381 --> 00:05:57.781 Florian, I'm still working with both. 00:05:58.281 --> 00:06:02.861 Florian is my co-founder at OmniFact. Alex is also part of the founding team. 00:06:04.481 --> 00:06:10.961 And from meeting these two, we got into building Rocketloop. 00:06:10.961 --> 00:06:16.861 Which was like a boutique for building digital products for well-funded startups, 00:06:18.101 --> 00:06:23.821 and larger organizations and I think the past seven or eight years we mostly 00:06:23.821 --> 00:06:27.421 built, first of all it was called data science, then it was called machine learning 00:06:27.421 --> 00:06:31.041 now it's called AI but basically state-of-the-art. 00:06:32.821 --> 00:06:39.221 Machine learning AI implementations but on a project basis we didn't have a product at this time, 00:06:40.171 --> 00:06:43.451 How? So Rocketloop is still around the company. 00:06:43.671 --> 00:06:47.191 You spoke about it in the past, but Rocketloop is still around and you're still 00:06:47.191 --> 00:06:48.891 working there as the CEO. 00:06:49.371 --> 00:06:56.511 What was kind of the kickoff, the spark that made you start OmniFact? 00:06:56.791 --> 00:06:59.771 Good question. There were two sparks, actually. 00:07:00.151 --> 00:07:05.871 First of all, Florian and me, we always wanted to transit from being a project 00:07:05.871 --> 00:07:07.691 company to a product company. 00:07:10.311 --> 00:07:16.431 And when GPT-3 came out, it was a little bit before ChatGPT Florian showed it 00:07:16.431 --> 00:07:21.951 to me and I was like okay, we can basically forget that we're going to do individual 00:07:21.951 --> 00:07:22.991 natural language processing, 00:07:25.031 --> 00:07:29.811 projects in the next few years so we really have to get thinking and have to 00:07:29.811 --> 00:07:34.651 find a way around that and at this point it was more like being sure that Rockloop 00:07:34.651 --> 00:07:37.431 is okay rather than actively searching for a product, 00:07:39.411 --> 00:07:43.191 And then we went to our customers and, as you know, we are located in Frankfurt. 00:07:43.491 --> 00:07:46.751 So most of our customers are in highly regulated interest industries. 00:07:47.191 --> 00:07:51.411 We work with banks, insurance companies, mad tech companies, pharma companies. 00:07:51.631 --> 00:07:57.511 All of these have like very high standards when it comes to where they can put their data, 00:07:57.851 --> 00:08:07.671 how they can use their data in some context, how they have to be able to get 00:08:07.671 --> 00:08:11.171 be conscious about data privacy and so on. 00:08:11.371 --> 00:08:15.051 And we went to them, like, a little bit provocative and said, 00:08:15.151 --> 00:08:19.051 like, look at this. We have a cool demo with GPT-3 at this time. 00:08:20.711 --> 00:08:24.331 When are we going to automate your customer service based on that? 00:08:24.511 --> 00:08:29.831 And then, first of all, they were all very excited. We did some prototypes. 00:08:30.111 --> 00:08:34.811 But then we talked to the compliance and information security people, 00:08:34.851 --> 00:08:36.951 and they were, like, very not so happy. 00:08:37.131 --> 00:08:41.451 Having a single supplier risk with OpenAI, sending everything to the US, 00:08:41.591 --> 00:08:43.691 was just already for prime time. 00:08:45.071 --> 00:08:51.051 And that was, first of all, a little bit of that was basically the start of Flora and me thinking, 00:08:51.211 --> 00:08:56.611 oh, maybe there is a chance to build a product that enables organizations to 00:08:56.611 --> 00:09:00.251 use these kind of technologies to automate their daily business processes. 00:09:00.691 --> 00:09:05.891 I see. So let's talk a little bit about OmniFact. We already know it's... 00:09:07.538 --> 00:09:10.878 What natural language processing model is behind it? 00:09:11.658 --> 00:09:18.698 So OmniFact more or less. We see OmniFact as a platform that enables AI adoption across companies. 00:09:18.918 --> 00:09:26.138 And allow me to circle a little bit back. So I think we all have heard so much about AI. 00:09:26.438 --> 00:09:30.498 And there's like McKinsey had a study in May where he said we can automate up 00:09:30.498 --> 00:09:33.458 to 30% of all business processes in Europe with AI. 00:09:34.158 --> 00:09:40.778 But as you know, or as everyone sees, there's very little to show for it right now. 00:09:41.098 --> 00:09:46.098 There's Klana, who has one good example how they automated their customer service. 00:09:46.218 --> 00:09:50.338 But beyond that, there are not many cases where we actually have great AI solutions 00:09:50.338 --> 00:09:51.698 in production right now. 00:09:51.858 --> 00:09:56.898 I actually have to admit, I'm currently in a little bit of discussion with Google 00:09:56.898 --> 00:10:00.238 Ads, and it appears they only have chatbots there anymore. 00:10:00.338 --> 00:10:04.598 And they reply always with the same stuff. But I'm a special case because I'm 00:10:04.598 --> 00:10:09.738 a freelancer working on the brand of Startup Radio, which completely throws 00:10:09.738 --> 00:10:12.818 out the rules, even though it's very legal here in Germany. 00:10:13.018 --> 00:10:20.958 And therefore, they are doing nothing else than repeatedly sending me always the same message. 00:10:21.178 --> 00:10:26.018 Yeah, do it with an exact A has to match exactly B. 00:10:26.198 --> 00:10:29.258 I can't do it. Yes, do it again. 00:10:29.538 --> 00:10:32.458 A has to match exactly B. I can't do that. 00:10:33.038 --> 00:10:40.498 So you always get the same message back, and it feels like you're bumping your hand on the wall. 00:10:40.618 --> 00:10:46.698 So there's good customer service with AI, and there's very bad customer service with AI. 00:10:46.838 --> 00:10:50.598 Well, I have to admit, I expected more from Google. 00:10:51.338 --> 00:10:57.258 I think, to be fair, Google is a large organization, and I think they were prohibited 00:10:57.258 --> 00:10:59.738 to use Gemini for a long time in their processes. 00:11:00.378 --> 00:11:02.718 So I think there's a good chance that this will change soon. 00:11:02.878 --> 00:11:04.418 But to circle back to your question, 00:11:04.698 --> 00:11:11.978 we try to solve the issues that prevent companies from adopting AI and building 00:11:11.978 --> 00:11:15.058 solutions like good customer services, 00:11:15.298 --> 00:11:19.758 really individual, has access to the CRM, to the product information, and so on. 00:11:19.978 --> 00:11:24.218 And you can bundle that to give you a very personalized experience. 00:11:25.238 --> 00:11:32.578 But to do so, we actually think it's a bad idea to focus on only one LM provider. 00:11:32.838 --> 00:11:37.898 So what we do is basically we have the possibilities to work with all cloud 00:11:37.898 --> 00:11:43.238 providers, so Entropic, Google, OpenAI, and all Azure-based models. 00:11:43.598 --> 00:11:50.518 But we also support self-hosted models. And that actually gives us quite a bit 00:11:50.518 --> 00:11:56.238 of flexibility to basically choose the right model for a job. 00:11:56.338 --> 00:12:01.198 Because, of course, if you want to write marketing content for startup radio, 00:12:01.518 --> 00:12:04.338 you probably want to have access to the most powerful model. 00:12:04.458 --> 00:12:07.078 And you don't have to think about data privacy in any way. 00:12:07.791 --> 00:12:11.791 Yes, because apparently the podcast is meant to be published, right? 00:12:12.331 --> 00:12:17.751 Absolutely. Absolutely. So if you work in this space, you could probably also 00:12:17.751 --> 00:12:19.911 work with Claw directly or JetGPT. 00:12:20.271 --> 00:12:25.711 But if you want to review job applications, for example, this is not the case. 00:12:25.851 --> 00:12:29.791 There you have to be very sensitive when it comes to data privacy and how you process that. 00:12:30.911 --> 00:12:35.091 And if you're a large organization, you probably want to use a private model 00:12:35.091 --> 00:12:37.691 that you host yourself, maybe, just for these use cases. 00:12:37.791 --> 00:12:43.751 And what OmniFact allows you is basically to set up, we call it spaces, 00:12:44.251 --> 00:12:50.951 like AI assistants that can automate certain processes and basically tie them to certain models. 00:12:51.091 --> 00:12:58.351 So you can basically set the data privacy restrictions based on the use case you have. 00:12:59.271 --> 00:13:04.551 And additionally, you can then connect your internal data and services also to these use cases. 00:13:04.551 --> 00:13:11.191 So you can make sure, okay, we have our block database and that is used to create 00:13:11.191 --> 00:13:15.211 a new block entry, but that can be used with the best public model available. 00:13:15.491 --> 00:13:21.771 But our HR space is restricted to our local LM that we run ourselves or we have 00:13:21.771 --> 00:13:23.591 in our virtual private cloud. 00:13:24.931 --> 00:13:30.891 But that can also interact with our personio and our database of applicants, 00:13:31.111 --> 00:13:33.951 for example. So that is the main idea. 00:13:34.851 --> 00:13:40.711 Of our platform to basically cater to the needs of organizations, 00:13:41.311 --> 00:13:47.291 in a way that we allow for every use case to have a conscious decision about 00:13:47.291 --> 00:13:48.851 the data privacy requirements, 00:13:49.071 --> 00:13:53.271 where you want to run information, and which of your systems and data sources 00:13:53.271 --> 00:13:58.651 you may connect to this use case to make it more efficient to allow for more automation. I see. 00:13:59.151 --> 00:14:05.971 When we talked before, you said that is three highly regulated products. 00:14:06.414 --> 00:14:12.434 Industries. And they have the usual headaches, control over the data, 00:14:12.654 --> 00:14:14.634 including data secrets and GDPR, 00:14:14.974 --> 00:14:23.934 connecting with proprietary systems and fast development in the new models and 00:14:23.934 --> 00:14:27.314 overtake actually the projects. 00:14:27.314 --> 00:14:30.894 Because OpenAI can much better, 00:14:31.154 --> 00:14:37.854 has more financial means to really work on an AI model than any given company 00:14:37.854 --> 00:14:44.034 or most companies out there who are willing to do that. Absolutely. Absolutely. 00:14:46.034 --> 00:14:50.514 You basically went down the list that there's actually now a pitch deck. 00:14:50.734 --> 00:14:53.174 So it's data control and security. 00:14:53.514 --> 00:14:58.554 Looks like I've seen it. If you have that established, you can basically have 00:14:58.554 --> 00:15:03.854 your data and services integrated into some AI solution. 00:15:04.034 --> 00:15:07.774 And that allows you to actually build meaningful automation steps. 00:15:07.954 --> 00:15:13.214 But you are absolutely right that if you would build our own models, 00:15:13.214 --> 00:15:17.394 it would be a David against Goliath case, but not one you could win because 00:15:17.394 --> 00:15:21.014 that's a play of power and money right now. 00:15:21.734 --> 00:15:27.454 But the interesting thing is you have a few certain developments that are super interesting. 00:15:27.634 --> 00:15:33.434 First of all, you have companies like Mistral, which do an extremely great job 00:15:33.434 --> 00:15:40.114 to build models here in Europe that are also available, hosted in Europe by a European entity. 00:15:40.114 --> 00:15:42.954 So you have the GDPR thing covered. 00:15:43.694 --> 00:15:48.414 And they also allow large organizations to license their models and run the 00:15:48.414 --> 00:15:49.914 large models within the infrastructure. 00:15:49.914 --> 00:15:56.114 And in addition you of course have Meta which released the Lama 3.1 models which 00:15:56.114 --> 00:16:01.674 are exceptionally well and also run within the company's infrastructure. 00:16:03.694 --> 00:16:09.334 Looking forward I wouldn't be surprised if in the future a lot of like meat 00:16:09.334 --> 00:16:14.854 sized models that can run on hardware that costs I don't know like between 15 00:16:14.854 --> 00:16:22.414 and 20k to buy could solve very meaningful use cases And once you have that established, 00:16:22.854 --> 00:16:27.914 I think the question will always be, do I need the largest, most expansive, 00:16:28.254 --> 00:16:35.414 but also most closed and single provider restricted model for the use case I have? 00:16:35.614 --> 00:16:37.654 Or isn't the... 00:16:39.416 --> 00:16:46.096 Aren't most use cases solvable by much smaller, cheaper models that I can control myself. 00:16:46.456 --> 00:16:51.756 And our guess is the future will be hybrid between like models you run within 00:16:51.756 --> 00:16:57.856 your infrastructure and public models and maybe some things in between. 00:16:58.256 --> 00:17:03.876 I think it's still an open question how much fine tuning will play a role so 00:17:03.876 --> 00:17:07.176 that you have like a base model, like for example, 00:17:08.276 --> 00:17:13.656 Lama 3.1 and you want to be basically adapted to your use case or your language 00:17:13.656 --> 00:17:16.356 or your knowledge from within your organization. 00:17:17.016 --> 00:17:21.796 If that plays a large role, then again, you're going to have self-hausted models, 00:17:22.116 --> 00:17:23.936 which would be very important for us. 00:17:24.196 --> 00:17:30.716 Another way to include companies or organizations' information are ways like 00:17:30.716 --> 00:17:33.056 retrieval augmented generation, where you basically. 00:17:34.456 --> 00:17:38.856 Use the language capabilities of a model, but basically force it to base their 00:17:38.856 --> 00:17:44.296 answers based on information you provide them based on the query or the question user asks. 00:17:44.616 --> 00:17:49.356 And from what we have seen so far, it looks like that these solutions without 00:17:49.356 --> 00:17:56.696 fine-tuning, without individual training are probably good enough for 85, 90% of the use cases. 00:17:57.256 --> 00:18:01.536 And it will be very interesting to see how this will progress as the models 00:18:01.536 --> 00:18:07.016 improve. Because I think right now, no one has a good grasp of how great the 00:18:07.016 --> 00:18:09.096 GPT-5 grade models will be. 00:18:09.716 --> 00:18:15.856 That is an interesting question. We have now doing a little bit philosophy. 00:18:16.436 --> 00:18:20.836 Can we get really down to business? 00:18:21.416 --> 00:18:29.476 What could a potential client that is listening right now actually do with their tools? 00:18:29.616 --> 00:18:35.496 Where can you apply it? where do you have use cases where you can actually use the tools. 00:18:36.590 --> 00:18:44.550 Absolutely. Absolutely. So the very first use case that works off the shelf 00:18:44.550 --> 00:18:50.750 and is very helpful for most organizations is basically knowledge management. 00:18:50.750 --> 00:18:57.930 Since you can use OmniFact to make the information in your document management, 00:18:58.210 --> 00:19:02.430 in your Confluence, in your Notion, and so on, systems available through an 00:19:02.430 --> 00:19:10.410 iAssist system that is centralized at one point, you can enable much better access to companies', 00:19:11.190 --> 00:19:13.150 information. 00:19:14.930 --> 00:19:20.230 Where we have seen this to be the most successful so far is within banks because 00:19:20.230 --> 00:19:23.950 they have something like, and it's hard to translate, schriftlich fixierte Ordnung. 00:19:24.150 --> 00:19:30.050 So they have to basically have a written version of every process that is relevant 00:19:30.050 --> 00:19:31.730 for the bank's organization. 00:19:31.990 --> 00:19:37.670 And they are normally organized by the department responsible for the process 00:19:37.670 --> 00:19:42.950 step, which leads to a document that has thousands, 00:19:43.330 --> 00:19:49.210 sometimes 10,000 of pages that are very hard to comprehend as a person going through. 00:19:49.470 --> 00:19:54.250 But with an AI system that can basically access the information and summarize 00:19:54.250 --> 00:19:55.370 it based on your question. 00:19:56.647 --> 00:20:03.727 We can make this much more useful as a source of information. 00:20:03.987 --> 00:20:09.907 And that's actually the first high-impact use case we have seen to actually 00:20:09.907 --> 00:20:13.867 make this process documentation that many organizations must have because of 00:20:13.867 --> 00:20:18.667 regulatory requirements to make them actually a powerful tool for their employees 00:20:18.667 --> 00:20:21.207 to access the information they need much faster. 00:20:21.507 --> 00:20:26.307 For everybody with no financial background, Schriftlich fixierte Ordnung literally 00:20:26.307 --> 00:20:29.147 translates to written fixed order, 00:20:29.407 --> 00:20:36.887 which is documentation implying a structured or formalized system. 00:20:37.287 --> 00:20:43.427 And this formalized system basically tells you how you have to behave your processes 00:20:43.427 --> 00:20:48.687 as an employee in a bank in order to be compliant with the regulations. 00:20:48.907 --> 00:20:53.047 Plus, it also can be audited by the supervisor. 00:20:57.327 --> 00:21:04.727 Plus, you can also be checked against this SFO if you're really doing your stuff. 00:21:04.947 --> 00:21:13.067 And believe me, if the auditor does such an audit, commonly known as a 54 audit, 00:21:13.247 --> 00:21:19.087 Paragraph 54 KBG Prüfung, it's completely a fun-free zone here. 00:21:19.087 --> 00:21:24.467 Believe me, that is the moment when every bank CEO goes, uh-oh. 00:21:26.107 --> 00:21:32.627 I want to share some context on that. 00:21:32.807 --> 00:21:38.727 Why this is so interesting and so hard. Because if you're not within the financial 00:21:38.727 --> 00:21:42.267 industry, having a documentation of process can't be so hard. 00:21:42.987 --> 00:21:48.847 But the interesting thing about this Schriftlich Fixierte Ordnung is you have 00:21:48.847 --> 00:21:53.467 many departments within a bank, and every department is responsible for their process step. 00:21:53.727 --> 00:22:01.227 So basically, if you want to describe the process of granting a loan to a customer, 00:22:01.467 --> 00:22:07.547 this will involve six, seven, eight, nine departments throughout the bank, one bank. 00:22:07.547 --> 00:22:11.367 And basically, you don't have one document that's grabbed the whole process, 00:22:11.447 --> 00:22:13.767 but you have one document for each process step. 00:22:14.706 --> 00:22:20.286 And that actually makes this documentation as it exists completely incomprehensible in many cases. 00:22:20.626 --> 00:22:23.966 But with the AI, you actually can now ask the questions like, 00:22:24.126 --> 00:22:27.906 how do we organize our loan granting process? 00:22:28.106 --> 00:22:31.366 Who's responsible for that? Who handles the risk assignment? 00:22:31.526 --> 00:22:34.926 How do we handle ESG risks in this process? 00:22:35.226 --> 00:22:40.366 And it will basically puzzle together the information from hundreds of documents 00:22:40.366 --> 00:22:45.766 to give you this explicit answer, which in the past would be very, 00:22:45.906 --> 00:22:51.206 very annoying to basically get in a very compact way. 00:22:52.186 --> 00:22:55.766 And you probably would have to speak to three or four people who give you the 00:22:55.766 --> 00:22:58.666 right hints to basically find that out for yourself. I see. 00:23:00.246 --> 00:23:04.486 So, for example, written fixed order. 00:23:04.746 --> 00:23:11.046 That's one use case. Can you tell us, like a few others, what you can also do? 00:23:11.146 --> 00:23:16.806 Because this one is, I would say, pretty much limited to European financial institutions. 00:23:17.866 --> 00:23:23.686 Also, we've worked with one of Germany's largest landlords. 00:23:24.566 --> 00:23:27.506 And they have an interesting challenge. 00:23:27.726 --> 00:23:33.186 They grew relatively fast and not only organic. So basically, 00:23:33.306 --> 00:23:37.906 they bought together a lot of buildings and joined or merged certain companies. 00:23:38.146 --> 00:23:42.786 And this led to the fact that for every building they have, they have a different 00:23:42.786 --> 00:23:46.386 kind of structure sometimes for the documentation about that building. 00:23:46.706 --> 00:23:54.406 So if you now want to ask, when was the elevator last audited? 00:23:54.586 --> 00:23:59.066 And was something found in the audit? Has it been fixed? you really have to 00:23:59.066 --> 00:24:02.486 dig into your document management system to find this kind of information. 00:24:02.686 --> 00:24:08.546 Sometimes through PDFs that are not, weren't run through OCR, 00:24:08.726 --> 00:24:09.926 so you can't really search for it. 00:24:10.806 --> 00:24:15.066 And we help them to basically build AI systems for these buildings that you 00:24:15.066 --> 00:24:16.406 now can ask these questions. 00:24:16.726 --> 00:24:24.086 And this basically reduces the time that they have to get this information from minutes to seconds. 00:24:24.706 --> 00:24:29.646 That is not a use case. I'm currently focusing a lot when I speak about the 00:24:29.646 --> 00:24:33.226 use cases on what we like, what is actually shipped right now, 00:24:33.346 --> 00:24:36.626 but to, to extend it a tiny little bit. 00:24:36.866 --> 00:24:41.546 So currently we focus a lot on like how we can make the information that the 00:24:41.546 --> 00:24:44.006 organization has and that's persistent. 00:24:44.186 --> 00:24:48.046 So something that is in confluence in the document benefit system to make that 00:24:48.046 --> 00:24:49.966 accessible through an AI system. 00:24:51.228 --> 00:24:54.088 But in parallel, we are currently working very hard on making, 00:24:54.508 --> 00:24:58.288 also giving our AI systems access to live systems. 00:24:58.528 --> 00:25:03.328 So if you ask the OmniFactor systems a question, the first thing it asks itself 00:25:03.328 --> 00:25:06.888 is, what kind of tools do I have to answer your question? 00:25:07.408 --> 00:25:11.748 So, and right now it has two tools. It either can give you an answer directly if it's trivial, 00:25:11.808 --> 00:25:16.848 or it can go and ask a knowledge base, which basically is a retrieval augmented 00:25:16.848 --> 00:25:20.728 generation implementation that basically gets you the information based on the 00:25:20.728 --> 00:25:22.788 data source of where we're at it. 00:25:23.028 --> 00:25:27.528 But what we currently do is we're going to provide a number of additional tools. 00:25:27.648 --> 00:25:31.348 One that I showed you a little bit before the call is we're going to have web 00:25:31.348 --> 00:25:34.048 search and web browsing, which most people know from JGBT. 00:25:34.308 --> 00:25:38.268 But what is more exciting is we're currently building something like a plug-in 00:25:38.268 --> 00:25:45.068 system that allows you to create a tool that, for example, would do API calls 00:25:45.068 --> 00:25:47.668 to your customer relation management system. 00:25:48.388 --> 00:25:51.028 Or to your product information system. 00:25:51.228 --> 00:25:57.288 So if you combine these tools, then you can automate very interesting things 00:25:57.288 --> 00:26:02.568 because then you can say, oh, I have this customer email and the customer complains about, 00:26:02.828 --> 00:26:08.228 I don't know, a broken piece of hardware that they bought from you. 00:26:08.228 --> 00:26:12.528 You can basically then take this information, look up in your CRM, 00:26:12.708 --> 00:26:15.188 see if the customer actually bought it, when they bought it. 00:26:15.308 --> 00:26:25.148 Then you can go and check your knowledge base for the warranty rules for this certain product, 00:26:25.348 --> 00:26:29.508 can check that against the time it was bought, and can basically check if the 00:26:29.508 --> 00:26:33.708 customer would be eligible for exchange, for example. 00:26:33.708 --> 00:26:37.708 And that you can basically now automate end-to-end because our AI system can 00:26:37.708 --> 00:26:39.168 basically plan the steps. 00:26:39.488 --> 00:26:43.068 It will ask you, hey, Joe, this is the steps I would like to go. 00:26:43.188 --> 00:26:47.748 This is what I want to do to provide you with an answer. Then you can say, no, I don't know. 00:26:49.568 --> 00:26:56.508 I know already it's beyond warranty, but there's a certain reason that we may 00:26:56.508 --> 00:27:01.528 exchange it anyways if it's from that product line, for example, 00:27:01.648 --> 00:27:03.068 because we had issues with that. 00:27:03.708 --> 00:27:08.088 Then it would change the plan, then it would run the plan and give you the complete answer. 00:27:08.468 --> 00:27:13.508 And these kind of like end-to-end automations that become possible once you 00:27:13.508 --> 00:27:17.668 have access to not only data, but also like systems within an organization. 00:27:18.928 --> 00:27:22.488 This is what excites us a lot, because I think at this point, 00:27:22.568 --> 00:27:24.788 you can make people much more efficient. 00:27:26.085 --> 00:27:29.885 For example, what would be great if at one point in the future, 00:27:30.225 --> 00:27:34.945 your sales employee meets somebody on a fair, talks to his phone, 00:27:35.065 --> 00:27:40.145 says, okay, offer this and that to this client, get it ready for me in the drafts. 00:27:40.265 --> 00:27:44.185 And then the next day he's in office, he just has to review it and click send. 00:27:44.725 --> 00:27:48.205 Absolutely. I have another example that comes from my daily work. 00:27:48.345 --> 00:27:52.705 So currently with all larger customers, I still have like one-to-one calls to 00:27:52.705 --> 00:27:55.145 basically understand their needs, understand their use cases. 00:27:55.885 --> 00:28:02.985 And really still learn, also doing sales, but also to learn for us what are interesting use cases. 00:28:03.705 --> 00:28:08.085 And after this call, I normally, I write down a lot. 00:28:08.245 --> 00:28:11.725 So after this call, I go to my CRM, basically put in what is interesting, 00:28:11.745 --> 00:28:16.265 if it's an interesting lead, how large the organization is, and all this information. 00:28:16.505 --> 00:28:20.605 I go to our Slack channel and report back a little bit because it's sometimes 00:28:20.605 --> 00:28:23.785 very exciting and interesting for our team to see what we're working on. 00:28:23.785 --> 00:28:28.785 Then I go to Notion and put in if we have any feature requests that are new 00:28:28.785 --> 00:28:32.985 to us that we didn't have, or if they mentioned a feature request or feature 00:28:32.985 --> 00:28:34.785 that we don't have yet and add that. 00:28:36.365 --> 00:28:41.045 And then I send them an email, depending on how the call was, 00:28:41.345 --> 00:28:45.225 with our product presentation and so on. 00:28:45.785 --> 00:28:51.965 And ideally, I would like to just put all the things I wrote down in OmniFact, 00:28:52.085 --> 00:28:54.025 and OmniFact would understand, oh, I see. 00:28:54.705 --> 00:28:59.285 This customer is interesting to us. It has this potential of seeds. 00:28:59.445 --> 00:29:02.145 They want to test with that. I'm going to put this into the CRM. 00:29:02.345 --> 00:29:05.825 They also mentioned this in these feature requests. We have that in Notion already. 00:29:05.905 --> 00:29:06.965 Let me update that for you. 00:29:07.185 --> 00:29:10.645 And also, here's a summary I would like to set to Slack, and is it okay for 00:29:10.645 --> 00:29:17.165 me to send this email to the customer? so and if we get this point like my life and the life of. 00:29:18.190 --> 00:29:22.210 Million of salespeople will be much, much easier. And you can basically think 00:29:22.210 --> 00:29:28.110 the same thing about marketing, about your work with setting up the interview with me, for example. 00:29:28.690 --> 00:29:32.990 There's so much potential for automation. If you have these tiny steps and this 00:29:32.990 --> 00:29:36.970 planning ahead that you still can interact and change with and nothing happens 00:29:36.970 --> 00:29:39.370 with you approving what the AI plans. 00:29:40.230 --> 00:29:43.930 But this is basically the midterm goal that we want to achieve. 00:29:43.930 --> 00:29:50.550 Allow for this to be the glue between all the systems and use AI to actually 00:29:50.550 --> 00:29:54.050 have like end-to-end automations that take away like the burdensome, 00:29:54.190 --> 00:29:59.110 annoying little steps we have to do by changing tools, changing platforms, 00:29:59.710 --> 00:30:01.550 duplicating data and so on. 00:30:01.710 --> 00:30:09.030 It also sounds to me like you would make the SaaS process more trackable as well as more efficient. 00:30:09.830 --> 00:30:13.450 Absolutely. Absolutely. Do you also see, 00:30:13.710 --> 00:30:18.610 because you have sales on one side, do you also see something like this you 00:30:18.610 --> 00:30:24.330 could offer to customers in terms of product development, product ideas that 00:30:24.330 --> 00:30:27.550 they may like to have in the future? 00:30:28.490 --> 00:30:30.630 That's a very interesting question. 00:30:31.630 --> 00:30:35.130 If you think from a product manager perspective, 00:30:35.610 --> 00:30:44.150 I think having better access to user feedback by being able to search through 00:30:44.150 --> 00:30:50.510 it more efficiently and also by being able to search the web, 00:30:50.770 --> 00:30:57.370 compare with the competitors, and to probably combine information from your 00:30:57.370 --> 00:31:00.390 CRM, from your support and so on. 00:31:00.390 --> 00:31:05.710 To get a more holistic view of your customer's actual needs can be very helpful. 00:31:06.190 --> 00:31:13.130 In addition, what's also important, if you have tools that have analytics within 00:31:13.130 --> 00:31:19.510 your app, you want to combine this information, like how often was feature X used versus feature Y? 00:31:19.870 --> 00:31:22.110 Can you group that into cohorts? 00:31:22.870 --> 00:31:27.310 These are all things that currently are done manually that maybe not now, 00:31:27.410 --> 00:31:30.110 but in six months or so, AI could do for you. 00:31:30.390 --> 00:31:34.850 And again, we see ourselves as a platform enabling that. 00:31:35.110 --> 00:31:39.850 We are not a CRM. We are not an analytics tool, but we want to integrate. 00:31:40.090 --> 00:31:44.510 We are integrating with these tools to basically tie this information together, 00:31:44.510 --> 00:31:51.790 to give you like an assistant, like a student intern that can do meaningful work. 00:31:51.790 --> 00:31:56.150 It probably needs some oversight and some correction from time to time, 00:31:56.290 --> 00:32:03.470 but it helps you to do complex steps with a few instructions and less amount 00:32:03.470 --> 00:32:07.850 of time spent than if you would have done it yourself. 00:32:08.945 --> 00:32:12.785 Sounds pretty good. So we already know about your future plans. 00:32:12.965 --> 00:32:17.105 We usually close out those interviews with two. Now we do three questions. 00:32:18.865 --> 00:32:23.125 I would assume you are open to talk to new investors. 00:32:23.885 --> 00:32:32.365 Absolutely. So when we transitioned from the project-based Rocket Loop work 00:32:32.365 --> 00:32:36.225 to our product company, we did a very small friends and family round. 00:32:37.065 --> 00:32:42.085 And we are currently planning for a pre-seed round, seed round. 00:32:43.525 --> 00:32:47.165 The pre-seed round probably going to close by the end of the year and the seed 00:32:47.165 --> 00:32:48.585 round somewhere next year. 00:32:48.745 --> 00:32:53.485 So we are very much interested into discussions with investors. 00:32:54.825 --> 00:32:59.405 There's still some room for business angels, but at the end of the year, 00:32:59.565 --> 00:33:02.105 we will probably talk more to institutional investors. 00:33:02.745 --> 00:33:09.845 Sounds pretty good. Would you also like to hire more talented people? Do you offer jobs? 00:33:10.425 --> 00:33:16.085 Yes, mostly in communications, marketing, sales, and engineering. 00:33:17.381 --> 00:33:21.601 Mm-hmm. For everybody who would like to learn more, for the investors, 00:33:21.881 --> 00:33:27.421 we link down your LinkedIn profile, and we also link down here your career website. 00:33:27.421 --> 00:33:32.821 My last question, since this is sponsored by Hassan Trade & Invest and the European 00:33:32.821 --> 00:33:39.181 Enterprise Network Hassan, is there something you would like to address to the 00:33:39.181 --> 00:33:40.921 decision makers here in the state, 00:33:41.201 --> 00:33:46.821 what they are already doing good and or what they could improve in order to 00:33:46.821 --> 00:33:49.101 make this a more sustainable startup hub here? 00:33:49.581 --> 00:33:54.701 The thing is, there's Hassan AI and they do a pretty good job. 00:33:56.501 --> 00:34:03.981 I think we have a very strong hop in Frankfurt or in Hessen when it comes to 00:34:03.981 --> 00:34:06.081 finance for the finance industry. 00:34:06.281 --> 00:34:10.861 We also have one with the pharma industry, but I have the impression that they 00:34:10.861 --> 00:34:19.481 are not so much involved with all the startup AI and data meetups and events and so on. 00:34:19.561 --> 00:34:25.061 So maybe that could be an improvement. what is interesting though we have been 00:34:25.061 --> 00:34:32.461 in talks with public entities but not from Hesia yet so, 00:34:33.881 --> 00:34:38.421 from what I know so far Hesia does a great job, 00:34:39.981 --> 00:34:41.921 with enabling startups, 00:34:43.641 --> 00:34:50.061 hopefully the next time we speak I can say more about the AI aspect of it because 00:34:50.061 --> 00:34:55.601 for now, we have not taken too much steps into finding that out. Sounds pretty good. 00:34:56.881 --> 00:35:01.901 After long weeks of trying to organize that and more than 35 minutes of recording 00:35:01.901 --> 00:35:07.061 time, Patrick, I wish you all the best, best of luck, and thank you very much. 00:35:07.361 --> 00:35:10.361 Thank you, Joe. It was a pleasure, and thank you for your time. 00:35:10.841 --> 00:35:12.961 Bye-bye. Bye-bye. 00:35:16.880 --> 00:35:43.232 Music.