let me maybe start after after the of the talk from from from tack which I very much liked is know everybody may be scared of China but I can tell you our motivator a little bit now on this on this last presentation yeah so we don't need to be fully scared of China so in just couple of stats we have China has I think last year produced 20 unicorns Europe has produced 17 unicorn so we're not too far away in Europe in Europe we did this with 22 billion of investment in China with 60 billion or almost 70 billion so it's not that we are we are turn as much bigger and producing these papers and all these kind of things so but let's not give up let's try to try to build something here and we can build something here another point may data point talk mention mentioned sense time which has been the most valuable AI company globally now we have a company from Europe which is the most valuable a company called
uipath and such as the most valuable yeah i company coming from romania in this case but it's also the fastest growing software company ever so we've never seen a company growing that fast and in the enterprise software up space so coming from europe so there's not this hope so if it's this sort of before I start with this one here um just to let you know that there's there's hoping we see big potential also in AI although I must say we always say as early bird and we say we don't invest in AI because you can't eat you can't do anything with it we also don't invest in blockchain but if we're looking at our portfolio we have 13 what's a deep AI based companies would you really use AI for the product from may for differentiate with a product she take a little portly probably 15 to 20 which is it is almost a third of our extra portfolios we have quite some AI companies in our portfolio but we don't talk too much
about it because again you can't eat AI and we rather look for business models and this will also be part of the story here what are we looking for in the end to make make investments and and hopefully find more uipath by the way our portfolio companies which we were very happy about we can imagine and it just is that so years ago the company launched the product they are now at the run rate of almost 400 million arr they are just valued at 7 billion so just before the six months ago they were valued at at almost three billion so it shows a little bit this kind of dynamic is possible in Europe it's not impossible it's doable so we have a lot of AI and our portfolio we see a lot of AI basically if we look into our deal flow how we call it so the investment overages we see we have clearly one third which claims at least having AI if we would drill this down further in terms of which
companies are we more deeper looking into them is probably hundred percent so in the end no one comes comes through our door without and saying I have AI or blockchain something this is just you know something everybody has that's why it's confusing it's you don't know exactly what it is because everybody claims I have AI and I have something very cool and which nobody else has but over it is not the case and for us it's not so easy always to distinguish between what is real AI what really makes sense what really helps makes business models better what really distinguishes in the end what makes what makes the customer experience better or whatever so it's confusing it's it's also the older boys the question how you look into things because things can don't look very very different if you look at from the from which perspective you look at it so some may really claim they have great AI if you then talk to a professor who says by no means this is AI this
is far away you can't even you know touch the word AI with what they are doing so it's very very different also in terms of what what's the use what the use cases are so that's why we why we also try to basically look a little bit deeper into this and try to try to make make more sense and try to basically structure it a little bit so in the beginning and I'll go through this quickly in the beginning there was just the descriptive basically things so this was was the beginning something obviously which was interesting at that point in time but probably not any anymore so if somebody comes with this before II not so interesting the next thing was is kind of the thing which is more in the terms of predictive so what might happen that's obviously more interesting and there's also many many of the companies we have on our portfolio exactly doing this but this goes further and and you know this in terms of tell you what you need
to do have the really coming up with suggestion or even automate things in terms of automatically things are happening and that's obviously I wouldn't say that's the Holy Grail that we will see probably a next step after this but at least for now I would say well this is obviously something we are more interested in and if companies are approaching us and there were we have something in this direction we per se find this interesting and I'd like to talk to them you'd like to take deeper because that's obviously what we want we don't want basically just a better in terms of what might happen prediction but we want went in the ideally we want basically the algorithm take these decisions on on its own but self-driving obviously is going in this direction but you all know it's not working as of now at least not to talk to mass adoption but we are we are on the way to get there and this is obviously what we also want to see with our applications
not just self-driving applications we have basically two we see two general differentiations we see the the one which is the horizontal approach and the vertical approach come to this later and they are a little bit different and so what is the chorus and this is the value change you see here of AI and we see many many players basically being here and not covering just one part of the value chain but but but some of them some some even do all but have certain strengths in one of these sectors and basically elements of the of the value chain I think that's that's clearly interesting some of these companies we are we invested some of we have seen some we liked very much but so this is one way the other is and that's quite obvious that's the more vertical approach in terms of becoming the best in one in one industry basically and and applying the value chain the the the AI value chain in one one sector and to make it a to exciting
in the end for us both cases can work both cases can create unicorns both cases can can have a great impact and because we've seen this in the past they're examples of companies who basically got there and and we see more who can basically become become this become relevant companies the question now is when when is this something when when when is there relevance just an example of what the companies do phones and frogs sir they do just just they apply AI form for payment so base do to detect fraud in payment so just that just very very focused but it's a huge market it's a global market is a huge market basically with with a tremendous impact basically can disrupt industries you can disrupt part of the payment infrastructures with this we can disrupt part of the CAD schemes and so on so it has a huge impact basically with just one application or an application in a vertical and there are many many other examples of companies who can who can do the
same if I flip back to this one here obviously I haven't mentioned uipath here Europe has is probably more than home or horizontal approach can be applied to any industry basically they claim is even that they say well everyone need a UI engine on the desktop what they do is maybe you know everybody knows what they do they do robotic process automation so in the end they take over a repetitive task you you're doing and everybody does this every day so they're taking over this they check what you're doing and the day take over and take over these tasks so in the end their claim now is everybody needs it so everybody should have a this engine on the on the desktop so there's some logic to this and obviously the more applications they have the better the better they get so this one example of a more horizontal approach where you can where you can basically have used it for yeah it said as they claim now as they want to where they want
to go to have everyone using their algorithms what what is very important and this goes for both approaches and I come to this when finishing in terms of where do we how do distinguish between the two general approaches vertical horizontal what is very important is the point of the login so and the good thing is with AI not in many cases or many applications you can have you have good luck and you can create a login for instance with network effects like Facebook you haven't you don't you don't they will say well we wanted to leave Facebook but nobody's doing this because you have these network effects and they're very strong with data you can create the same because if you have trained your engines on on data you can basically create look and effects because the company's being behind they cannot do this because they obviously have to catch up and since you're generating further more data and train your engine better over time you become and stay stay ahead so Percy with applying
a eye to something using eye you can create these locking effects which we as a VC like very much so this is exactly what we're looking for and we're not looking for just for these locking effects in terms of people and what what do we mean with this people not not switching so your customers not switching toward the next customer that's one that's the be called backward locking lock and effect so that you don't basically don't not being designed out basically but even more interesting is the forward-looking locking effect and this you can can achieve with data what that means is that basically the next customer you approach they have to take your product because they are so good and what they when they do in the data generation and in the end what they have built that they have to choose you because you have basically processed most the most data and much more than your competitor did and that's why you basically that's why the previous prima choice always and this kind of
forward-looking lock and effect is is amazing and it applies for some of our companies here for instance I mentioned Rockstar in both cases so in the vertical and the horizontal for for your iPad it's a given because they obviously learn was more task and they get more specific even in vertical so they can apply in all verticals because because they're they're so broad but also in the Fox the case basically they get much better over time the more data the more basically payment and they payment volume they process every day so and then they can't obviously like you can't reach them at a certain point in the state in time because they've processed so much data that they always give the better recommendations or basically it's even automatic so they do it on their own in terms of turning payments down or not so they're already at this stage in terms of terms of basically acting from from from not just recommending something but also acting according to that you can still find fine tune
and tweak this and set rules but in the end they can they can they can run it on do it on their own but this creates locking effects and this forward locking lockers effect which is which is very strong and and this is exactly what we would like to see and in in when making decisions and deciding on investments and as you can see here it's basically it's it's it's it's getting better over time so and you have this upward spiral which is which is very very attractive the issue is a little bit here that you have this chicken egg problem so the question is what does she have to basically build both so in this circle basically obviously the more data you have the better you get but how do you get more data you need more customers so but how do you get more customers if you don't have enough data or process data so this is exactly the issue which needs to be overcome and that's that's that's a given but and
there's no recipe basically to overcome this there are some some ways of getting there I come to this animated but but that's that's that's an issue but obviously this is also the the flipside is that can be very strong once you're on this path but it's obviously you have to get there that's that's the biggest issue um so in a nutshell pacing coming coming a little bit to the end of what we look at and what we do in that so we like both so this is here showing the horizontal and the vertical not part of the nut what is it what what what is important basically what we would look at if we would see the horizontal or the vertical approach I think you have to be really good and in very and if you have a horizontal approach basically you have to be cover all the the parts of the value chain in a very basically intense and deep way so you have to be really good across the entire value chain you
also have to be very close to research that that's important you have to basically try to get a way to generate data in a certain way so you need to basically get access to a data you have to basically find find an approach that gives you the data even by even if you need to give way can share the data with others so if you for instance work for one customer many would require that the data stays in in their house even premise has to be done in many cases you need to find ways of getting getting access to this data they even and use this data make use of this data even if you if you if you're serving other customers I think this is kind of the Holy Grail but this is really really important and that's that's an issue because many would say well we don't we don't we don't like this but then unite need to find other approaches like that machine learning or so on so you need to basically
come with with ideas of still getting the essence of the data of the customer and make use of this train your model with this and just share the model information then with with others before with your engine better to make it better I think these approaches are really important too and you can't overcome with this the issue of that customers don't want the data to leave their their their house and because they claim it's their own data all you find models were basically new customers benefit from the data from other customers we say also seen have seen these kind of models where you basically can can convince them to share part all parts of their data because they have a you big they get benefit from for the other day other competitive competitors suppliers whatever data they can use even if it's just parts of this we have a company in our portfolio saying they use both basically they do exactly that and share this kind of data with use with combining AI and blockchain
was to Boswell's combined so it's very very nice so they're they're doing these kind of things yeah and probably have read through this and now here but but in the end the essence is both approaches can work we like both basically we don't have a preference for one or the other it's just met us that you have can create this lock and lock and effect spacing that you can show early on obviously if you're far down the road and have all this early said you don't have this but we still like early stage investments and the question is how do can you overcome this but I said there there are ways to overcome this situation that you have don't have the data and can share this data with what I've seen and we would like to see this in an early stage that you basically have these ideas of how to overcome this and can show even in small batches or small data already a some AI working in work in place so if somebody
comes to us and say well we have this we have all this data but we need even more data to make sense of the data and obviously we have the credit but why is this soaked why can't you get access to this data I can at least show us that it's that the engine is somehow working and we always want to see this engine working we see in many many cases we see we see companies which claim they have AI but if you look deeper there's hardly anything so there's hardly any there's a kind of rule-based space system where the AI needs to be built and we really try to also and this is also advice in terms of that you want this you need to hire the right people obviously who have the dead did the DNA of really building this and from the first day on because you design obviously your product on the first day on according didn't what the needs are and according to basically really applying these AI algorithms from
the first day on yeah that's it from my side yeah we have a bit of time for questions I guess okay Christian so let's start here in the front that was a great presentation Jesus in in the sort of race for AI supremacy this there's a whole fight going on about who's gonna have the best chips to run AI who's gonna have the best cloud services who's gonna have the best ATP is as a service for vision speech sound that sort of thing do you think role Europe has a role play in those levels of this or the AI stack um yeah well this is something where we see well it's always it's it's it's floating so it goes with hardware and so on but I think it's the we see development you basically you can't we wouldn't invest in this type of thing because you basically the next generation will come and the next vibration will be will be better and in the in a couple of dimensions so that's why I think we
would probably be more reluctant and in investing in something where you may have an edge now but where you already know that the next generations kiss is coming soon and do you think Europe though broadly can compete with the US and China and in these areas yeah well has said in the beginning I I believe so and we see at least as our portfolio we see more and more interesting tech companies up we see this and the investments we will look at we we look more intensely we see this pretty broadly and and in the end the stats show that these companies can be created here obviously we don't have the the support in terms of the financing compared to China but obviously we do it somehow more efficiently here is it go Tibet no it's not god I would rather see much more money flow flowing into the end of the system here this is something which is known to everyone it's it's a factor of five or eight between between the US and
here we need to so this is clearly clearly not positive but we need to create more of these of these cases in terms of these promoters or cases I think when we see this happening more more money will also flow into the ecosystem great do we have another question for Christian that's the last chance yes Thomas cause speaking would you invest into an endeavor of a general purpose intelligence already now or would you wait a couple of years to do so well very broad question I would say it depends I said we would we would invest in horizontal approaches but it really depends then and it really depends how advanced is the technology it is there really an edge on the technology side it has to it has to be otherwise you have no chance for the horizontal approach all right final question there's one thank you very much for the talk and how is actually use of data effect that you as your in as a VC in your daily operations whether that's in
sourcing or in evaluating investments it's a good question we have actually made up our minds whether we can where we can basically use more AI to to for pattern recognition or for detection of any any signals in terms of is this could this become a good investment or not cut this become profitable we're actually in deep thinking about whether whether we should do more in this specific sorry to man with respect or whether we should basically dig and then build a model for that we are very yet but we have discussions going on basically of what and we have whether there's some what as mentioned we have our our module based system and in that sense but a little bit it is like this can we automate this I'm sure we could could be totally automate this and the sewers of now you