and our next speaker will also talk about strategies but not for governments i presume i think he has not founded a country yet but i can't be sure because before co-founding morentek's the world's first ai venture studio here in berlin he sold his previous company for a hundred million euros and that would buy you a little banana republic i presume i'm not sure um please welcome a serial entrepreneur that has started more than 10 companies in europe and the us in a wide range of fields ariane loja give it up for ariane so thanks a lot thank you very much for this warm welcome and uh i have to say probably as for most of us it's a premiere for me to have this kind of format um i'm looking forward and when fabian invited me to talk about strategy well my first thought was this is exactly the thing i'm not going to talk about today but i will also tell you why because i think we have done a lot of strategies in
the past especially when it comes to ai and i was following uh some of the conversations on stage here uh before and a lot of the notion was yeah okay so we did a lot of planning we have a lot of ideas but we fail in actually getting it done so this is what i would like to talk about today so i'm not going to talk about strategy but about actually building something and what it takes to do so so let's have a look at a few numbers that i think are pretty interesting because i think we have a lot of ingredients here we just fail short in actually putting them together so let's look at um researchers for example in the space of ai if we count them on a per capita base actually germany and and most probably as a good example for europe has more researchers than us and china together however we fail short in transfer which a good idea is if you look at the investments that happen in those
spaces um in this space of ai so if you think about germany and you've listened to a lot of the talks uh this morning uh you also get a pretty good idea we have actually really really great research in germany in europe in fact we have the top research institutions in ai we have amazing companies we have amazing industry players we are probably the best in doing hardware um we have um won most of manufacturing and we still are um really really good in that we have a lot of companies who are world leading in their field and they all have everything that it actually takes to get to the next stage still a lot of them don't make the transfer from being a great hardware player to becoming a great software player on the talent side i think also again based on all our research institutions we have access to the best talent that that you have however also there we somehow fail short in assembling these elements and putting things together um and
i'll come to that in a bit uh how we can solve that but first of all let me tell you why we believe ai is the most important technology in the 21st century at marantix as adventure studio so far we have built five companies in the medical field in the biotech field in the data management data infrastructure field and calls reasoning field which is a bi company we are about to build two three new companies next year our fund strategy is to build 10 companies in the next four years we are assembling teams we're finding solutions we're validating ideas we are creating the traction and then building a product that actually has a a great market value and is viable in the market so we're not only planning and strategizing but we're actually doing and this doing part is what i'd like to talk about in the next couple minutes so let's look at what actually are the important problems if you look at if you look at ai solutions so first of all you
need a ton of data and usually a ton of data is not even enough because the quality of the data is actually quite important so we need to assemble a lot of different data points on the same subject in order to get a full picture and usually what makes life so hard are the edge cases in building successful ai applications and those edge cases actually they require very specific data points that usually you don't find out there so you first need to create it so that gives you some scaling problems especially in the beginning then we have actually research that goes very fast so right now i think you know if you look at if you look at natural language processing for example two or three years ago most of the stuff that we look at today still didn't work yet today the whole world has changed and most people talk about language in a way like vision and image uh happened about eight years ago when we had this uh famous imagenet google moment
when actually vision was almost solved and we're now looking into language getting solved very rapidly so there's every month there are new um breakthroughs in in science and research that are actually very relevant for for um for the application side however when you then look at the sales effort how you are selling ai solutions then you look at very lengthy b2b uh sales cycles where often you it takes like six to 12 months before you can actually uh get a client on board um so there's a big mismatch in that um also if you look at the technology behind all this the the the machine learning models we see them as kind of a general purpose technology so we don't believe the technology itself will provide a big enough mode to build successful long-term viable companies it's more what you actually do using this technology that enables you to build really sustainable business models so in a way when you look at machine learning environments and machine learning infrastructure a couple years ago you had
to invest a lot of time a lot of engineering resources to build your basic infrastructure so that you could start train your models now today you have tensorflow you have pytorch and many many other frameworks that are just commoditizing this space so before that there was still a niche for companies providing this infrastructure and that's just an example how fast-paced this commoditization goes uh which in essence means this is not going to be an interesting uh model in the in the in the in a long-term perspective then another actually quite important part of what we're seeing when we build companies is you cannot get to 100 automation and everybody who sells that is actually failing and sometimes failing quite prominently so what i mean with this is if you look at our medical imaging firm avara the company in the in the in the uh radiology breast scan breast cancer early screening space there the company does automate 50 of the cases and the other 50 percent get routed to a doctor who still looks
at them and takes care of the more complex ones um of course every year the level of automation will grow bigger because our engine is actually um observing the doctors at their work and is learning from every single move a doctor does but still today it would be completely unrealistic to expect the machine taking over 100 and of course this human and loop is changing the economic situation of a business model and so you need to find smart ways how to factor that in and if you look at self-driving that's another example where it becomes very very obvious level five will not happen any soon and it still requires a human in the loop if we want to go to a level five automation now that of course gives some pressure from the investment side as well so if you look at how to invest in in an ai company it's an entirely different method than if you invest in consumer internet firms um so and that's also one of the reasons why u.s investors
um until today and also the chinese ones feel much more comfortable in investing in those moonshot ideas because they have been investing in those kind of businesses you know when you think about silicon valley the name says it all it has been growing in the space of silicon and silicon was always a billion investment business so you cannot build computer chips without investing billions and that just brings a total different attitude if you look at at investing so and here i come to my to my to my take take-home message what it actually takes to build successful ai products and also companies is you need to assemble the right ecosystem and also that's what we should be focused on here in germany here in europe to actually make sure we create the right seating ground right foundation to build successful ai applications so we need to bring those fields together to an actual ecosystem where we have very strong ties between the partners between the players where we make sure that information is traveling super
fast where we're making sure that also we're as we speak as we plan as we strategize we also start doing i think that's one of the main uh issues i have with our european culture you know when we think about uh self-driving cars for example the us culture is like okay once a car does less accidents than humans do that system is good to go in europe we're like yeah the system is good to go as soon as it does zero mistakes then it can go on the road well there's an obvious there's an obvious divergence here so let me talk a bit about the experience we have made at Merantix we have now been iterating our ecosystem over the last four years marantix was founded in 2016.
we're now a team of of almost 100 people um over over six companies as i said there will be 10 more companies in the next four years let me give you an uh some some some insight on on what we do and how we do it so first of all uh we are probably one of the most amazing talent machines when it comes to ai talents that's something i can say without blushing because i know we have one of the best machine learning teams in europe how we do that is actually we have built our reputation our brand our team in a way that today we are getting dozens of applications every single day and i have the luxury of looking at over 50 candidates per one hire i do and that makes a whole difference because i no longer uh only recruit the smartest but also the nicest and much more important of course the most the more the most socially skilled people who make really great teamwork which is in essence what you
need to have if you are working on these really difficult problems building ai solutions because one thing that is very clear is and that it's also where the the today system of of academia comes to a comes to a an issue all ai problems are multi-trans-disciplinary you need teams that work alike so you need to assemble teams with very diverse backgrounds with very very diverse languages and need to bring them together and make them work ai is all about know-how so we have built a lot of different formats inside our company how uh how know-how travels how know-how is transferred we have weekly sessions paper reading groups deep dives with a lot of external people also joining talking on very specific topics we were one of the founding partners of the german ai association which is also very prominently here today already seen vanessa and we make sure that you know it's not only about building businesses but also it's about moving science ahead and and and and again also here bringing the right people
together we also provide funding so we decided not to rely on external investors for our company only we have built three first companies from our own balance sheet over the first three years and we have raised our first funds or we have announced and raised our first fund just this year that is a 25 million fund backed by major u.s institutional investors that's of course also one of the things i wanted to highlight here yes it's u.s institutional investors and yes european use of u.s european institutional investors are still very hard to come by because they have a very different tradition on investing and they have a very different appetite for risk and reward of course because a lot of these u.s institutionals are so bullish for vc because they have made very very good experience over the last 40 50 years in actually getting returns that no other asset class was able to provide um last but not least it's all about also connecting industry and connecting um with the right players we do
not only build ventures but we also have built one of our portfolio companies who is very deep into germany's and european industry doing a lot of project work and exploring a lot of new fields also sometimes at the brink of of research and public funded research and uh and uh and and business um so we we at the same time usually run four or five european funded uh projects we just started a another project with uh with fraunhofer where um we're in with etiha we're in with max planck so there's a lot of things going on and we try to kind of concert this and also what is very important we do not try to own everything but we rather try to provide the right kind of base and also kind of an open source attitude so all the companies under our umbrella they are part of an open research agreement which lets them share everything with all other companies including data including contracts including whatever you can think of and there's a clear set
of guidelines and code of conduct what they can do with it and so i would like to conclude my speech with what this all leads to so so far as i said we have built a couple companies we all co-locate them under the same roof and just this year we decided okay let's give this whole thing a shot let's try to think our ecosystem at marantix like 10 times bigger and let's open it up for others to join it so we decided to actually plan and build our own campus which is going to be opening in april next year it's going to be on 5 000 square meters it's five minutes away from nordmanhof and it's gonna be uh yeah creating uh space for uh more than 450 people and and that probably are going to be 20 30 different teams um it's going to be startups but also corporate venture but also corporate machine learning teams research academia and also investors it's basically our vision to create this seating ground for ai in berlin
but also kind of acting as a as a role model for others and and my personal belief is that the way we do work in in machine learning um is a very good blueprint that can also be used for other deep tech spaces so um we would like to be very open there also to inspire others uh to build this kind of ecosystem for other deep tech areas which i believe is very very significant and important for us here um to make sure our prosperity is continuing a couple pictures but um most important go and have a look it's online and the building is there so you can also go and have a physical look it will be open in april 2021.
thank you very much thank you adrian and of course we all hope that we can when it's open i'll visit you without fear of covet or anything maybe this will be behind us next year so uh good luck with this we don't have too many questions from audience but i want to encourage you to still ask a lot of questions we have a couple i have tons more um maybe start with this one how can companies who aren't Merantix attract the resources like capital talent industry know-how etc etc i mean 49 people stay unhired when you pick one out of 50 right so can you repeat the question right how can companies who aren't Merantix attract the resources well it's all about the teams so it's all about the initial team so um the the reason why we have this factor today is how we started right so we started with a couple really crazy smart and nice guys and people in in a room and we have built our company around this we have
made sure that our hiring bar stays very very high all the time no matter how urgently we need people so we have never made compromises and we have tried to create the environment where the smartest people love to come and work with others which is usually it's not that hard right smart people want to work with other smart people and they want to have a nice environment they don't need too many perks so we're not a perk company i think the the campus will be really cool and and it will have some some kind of perks but it's not about the perks it's about the people uh you have in your team and making sure that um you have a clear set of values you use in in hiring right thank you um what are the strength of berlin for ai and maybe i want to add why not zurich why not the silicon valley why not beijing yeah yeah so i've been living in zurich before i've been living in san francisco and i've
deliberately chosen berlin um first of all because we have chosen europe um to build marantix so i was living in san francisco when we decided to build marantix so it was a deliberate decision to do it in europe and then we have looked at very different cities and we ended up choosing berlin the one major factor i believe is um we felt like berlin is a place where i can make people move to from all over the world um so that was a an assumption a hypothesis when we started in 2016 the background factor so today we can definitely we can definitely tick that uh we have in our team uh i think over 30 nations and 80 of the people working for us actually or one of our companies actually moved to berlin to work for us great and is that also like an asset for investors do they like to invest in berlin companies and come visit and yeah i think if you look at if you look at um if you look
at us investors who are so far investing a lot in the u.s and also silicon valley of course they see the bank for the buck that is just much better because my cost of living in berlin are just like i don't know a third or a fourth of what what costs of living are in san francisco and of course that directly translates into your cost base as a company uh where where usually salaries uh uh are are the most important factor right you were talking um about the uh the research and then the transfer right into uh into products um and i think you put it this way one is the ingredients and then you just you only have the job of putting it together and i want to challenge that i mean isn't that the exact problem like putting it together isn't that where uh you either make money or you lose money i mean in berlin you probably have noticed by now many people have a project not so many people start a
company right well you know i think um yeah we have all the ingredients and we failed to put them together and and of course putting them together is incredibly hard right so it's not that this is easy um but my my message is there is no reason why we need to stay and lag behind in comparison to the us or or even china um we have all the factors we just need to be a bit more bold in putting them together we need to um we need to risk more uh we need to change our attitude towards technology i think um not seeing it as an adversary but as as an enabler of of wealth over the last hundreds or even thousands of years and and embracing this and not stopping to push right and it's very interesting because i see a theme here today right i think in hans talk he was talking about like we're not hungry we're not really hungry in germany i think chris was talking about the lack of maybe
uh being wild and creative i was reminded of this old like um apple advertising slogan like stay hungry stay foolish i have the feeling we are neither in germany right do you really think that like one company or six companies under one roof can change this not now not now or not yet um but we're not we're just at the beginning here and we definitely have uh bigger visions for what we do and uh yeah you know maybe one of the one of the driving factors for rasters and myself we are the founders of of mirantix rasmus will be here tomorrow um i think one of our core drivers or key drivers is we want to do things we want to be role models to inspire others to follow us so instead of you know trying to convince everyone first we just move ahead we do it we get it done and we try to inspire others uh to do the same and i think in general uh as a society we need to embrace
the fact that um we are living in a very comfortable world here as a society and we have achieved a lot um however if i look at history in a grand scheme of things um all societies that have stopped being hungry and innovative they imploded at some point right and i would really hate to see us going that path right i i won't ask you where we are on the innovation doomsday clock yet but um ilka has a very good question and i must um i must disclose of course for all politicians what politicians watching this is now like free advice right free free consulting uh ilca's question is if you could make the decisions for germany's ai future what would be your next step i would make it a national priority first of all i would invest in research but also the transfer and the application i would also invest in the in the early education i think there we have a core issue of not understanding software and technology enough i think that's
two very easy ones for me at least i would also set up a lot of different investment vehicles with different strategies not trying to figure things out ahead but by actually doing them i think i would try to push much more on the doing side and also publicly embrace that making mistakes is okay and start to do so by setting examples by having government agencies fail more and make it public and also make it okay that it's like that i think that's a couple of ideas that i have top of my top of my mind very interesting i think if you if you don't uh fail in small details right then you're about to fail in the in the large scope which is probably what many people would uh would say about politics today very very interesting thank you i was very delighted that you said you can hire the nicest guys or that you are hiring the nicest guys because they're probably all qualified or overqualified this is something which is not very commonly
done i think in the tech world like there are people gurus like steve jobs famous for picking the a-list people but it was always the hard skills never the soft skills right so why do you do that and how did this change your whole way of working and your company you know i think when i need to solve a very hairy problem that can be solved by one person it doesn't really matter how nice this person is right how social this person is she he just needs to be incredibly brilliant and uh and talented now if i start solving hairy problems that are multidisciplinary i end up needing a lot of different people in the same room and i need collaboration to happen and i cannot have collaboration if people don't want to work with each other don't enjoy working with each other so you know i've been investing in companies for over 10 years now and everywhere where i'm i was losing money it was always the team failing first so it's not technology
it's not the market it's not the product that fails first it's always the team that at some point starts to break and then ripple effects the other things also fail so it's all about the team and i think my previous company was extremely successful and also much more successful than any other competitors in that space because we just had a better team and not only a more intelligent but a team that was collaborating much better at the brinks of the fields and combining things i'm not sure if you know or if people give you that feedback but do you have the impression that working with nicer people also makes you nicer maybe also makes you a nicer boss well i hope so i hope so i you know i i really uh try not to hire and i try also not to be an because i don't see a point in being an and i'm not saying that good leadership does not neces does not sometimes also involve making hard decisions and making hard decisions
will also be that some people will say well but i don't like this decision and i maybe even hate the decision but i think there is a big difference in you know uh disagreeing on on a topical level uh or having a fight on a personal level which i think is is unnecessary and also very unproductive right right thank you very much well i i hope that uh that is true i was afraid that people would probably not give you honest feedback uh but you seem quite nice to me as at least as long as we're standing on the same stage let me just ask you one final question you were telling us how you were building your ecosystem and how valuable uh and successful it was for you but you were hoping like like for an ecosystem explosion right you have something like 10 20 100 times bigger than what you offer what if that bigger larger ecosystem does not form around at least merchants in berlin what would you do like take the
company on the road or what is what is your idea for worst case if you have one worst case well you know i tend to not think about having a plan b uh i tend to try to uh make the plan a successful even though it might end up being plan a two or a three a plus yeah or a plus right so no i think essentially uh you know what we try to do is we try to connect a lot of people um also outside the ai world which is anyways happening very naturally because um usually we engage in in in different spaces as you probably have seen from our from our companies so um right now we're very keen for synthetic biology so we're doing a lot in that space we just start to look into quantum uh we start to look into neuroscience a bit more um and everywhere we go we try to bring new people together which is not necessarily always creating new business for us but uh we know
that in the long term it helps to you know build new kind of ecosystems thank you very much as a as a guy born and raised in berlin let me tell you welcome to berlin i'm really glad to have my antics i hope you will uh you will succeed in opening the doors having the doors wide open next april and i wish you a lot of success thank you very much thank you