# AI and Blockchain and how does it all fit together - Miljan Martic | Kosen LabsD2 V1 0023

- Channel: [ETH Belgrade Community](https://streameth.org/eth-belgrade-community)
- Date: 2023-10-07
- Duration: 39:55
- Watch: https://streameth.org/watch/yt-pLo44vb0oCM
- YouTube: https://www.youtube.com/watch?v=pLo44vb0oCM

## Transcript

foreign thank you so I must apologize first I think the title on the agenda is a bit different than the one here so that is the old title um I think the title was how does Ai and web3 come together but today I wanted to talk to you a bit about something more personal so it's about building at the intersection and essentially the lessons and a lot of the mistakes that we've made when we were building the startup over the last couple of years and hopefully you can learn something from that too so just a bit about myself first so I've been doing machine learning research for about the last 10 years I did my post-grad degree at UCL in London after that I worked in a number of startups around London and then moved to Google deepmind where I stayed for about five years doing all kinds of research primarily a Safety Research actually which is something that you hear a lot about in the news these days about AGI and Killer Robots and whatnot and then in 2021 I co-founded question labs which is a company that sits sort of at the intersection of AI and web3 and the idea was really to bring um to bring the kind of machine learning models and that kind of stuff into D5 primarily but then also to kind of wider space now when I accepted this talk actually back in February I was still the CEO of the of the company of course and Labs I'm not anymore actually I've left a couple of months ago um the company is still actually ongoing and my co-founder continued to to do the work we had some differences in Visions in sort of where the company should move so I decided to kind of pursue some other things um but yeah I wanted to kind of use this presentation to tell you about you know a lot of the stuff that we did and and some of the stuff that we we could have done better certainly so the way the talk is going to be structured is basically going to cover four main things where I think we've learned the most so first one is about fundraising so I think the previous talk by Timor was was a really good one and we'll cover some of the similar points there the second one is going to be hiring so we spend a lot of time on on hiring and finding with people and growing the team product obviously is is not an important one and the most fun is for the end it is about legal and Ops which even though it sounds dry it actually ends up being very very important so let's just Jump Right In um so in terms of fundraising just to give you a setup on what we did and how it went for us and it's it was actually a pretty crazy story I mean I didn't really know at the time because it was still my first startup so I thought that was normal but then later on when I talked to other entrepreneurs and friends they were like wow that's that's pretty crazy um so in June 2021 we started working on a ml prototype for essentially active liquidity Management on unusual V3 back then it was sort of a very new thing uni swap came back in came out in May 2021 so people didn't really know how to manage liquidity on these kind of concentrated liquidity exchanges and they were losing a bunch of money so what we wanted to do was create a machine learning model that kind of did this for you automatically and you could just put in the money and you would kind of fix these ranges over time as as needed in July 2021 we actually managed to create a prototype model um that that was working and had some pretty decent results the only problem was it was based on only like two or three months of data that existed at that point from unit swap and then in August 21 we started talking to a few friends there were VCS just to see if you can actually raise money we wanted to raise a few hundred K we were not very uh ambitious from from that point of view but then um you know things just started snowballing and we started to talk talking to a lot of VCS and and there was a lot of hype generated all of a sudden and you know on first of September I remember that day very well we actually had a call with A6 and Z where they gave us a term sheet um and they they gave us a five million investment which to us at back then was was insane so you know there we are from from kind of Zero to Hero in in three months had a lot of money and thought we were on top of the world so you know what are we gonna do um so in terms of learnings I think what we did well um first first thing we were really lucky I think it was the peak of the bull market and you know that makes things a lot easier um it's perfect timing the market was there The Narrative we had the narrative that was sort of crossing web3 and AI that was really good our background really helped obviously it was it was sort of a no-brainer I think one thing that we did really well is that we kind of tried to move really fast so in the month of August when we actually started talking to VCS we just said okay people are seemed to be interested let's just talk to as many VCS as we can in as little time as possible and what that sort of brought was that you know these pieces they talk between each other all the time and if you know a couple of them like you they'll tell their friends then all of a sudden you have this hype spreading out and what it turned out was that we didn't even contact A6 and Z they contacted us through some channels and they said hey we heard you're building this thing we're super interested which again came as a shock but it allows to basically play this game really well and then force them to give us a term sheet like really really quickly because we had interest from different sites and I think the idea itself was pretty good because it was anchored in sort of reality and what I mean by that is that it's tangible enough for people to understand for VCS to understand it's like okay you're building this ml model on top of unity everybody understands that but it still has a lot of prospects in the future you know people are expecting these amms to grow for the market size to grow so it has a potential to kind of develop into something interesting now a couple things that we didn't do well and that would definitely change this time around is that we signed a safe agreement and you know for those of you that don't know safe is it's basically a convertible that instrument it's very popular in sort of pre-seed and Seed stages and safe stands for simple agreement for future equity you basically agree on some terms such as in the future when you do a price round and you you actually issue Equity will give you a maximum amount of money that your cap that your company is going to worth uh anything above that will pay less right so that's all good and dandy but the safe agreement was actually invented in 2013. um and the problem with that is that it hasn't gone through a proper bear Market a proper Tech bear market so you've never people have never really seen what a safe agreement looks like when you have to do a big down round and it's just not pretty you know you could you could think that with a safe agreement you're selling sort of you know 10 of your company and then a bear Market comes when the actual safe converts the actual instrument converts into equity and you're down like 35 percent of equity which is definitely what you don't want to be um so the aim for us I think next time when we're doing this is to kind of do a price round straight away we were actually offered the option of doing a price chart on with A6 and Z but we didn't want to because it just takes a bit longer and there's like more lawyers involved and we just wanted to get the money too you know quickly um but I think that's uh that's a cost that you have to kind of calculate in and I think this is maybe a more controversial one but I think we also raised too much money so even though 5 million was sort of a standard raise back then three to five million was sort of standard in the bull market um I think raising too much money can dilute your focus and it can make you sort of try too many things at the same time and just make you too fat too quickly and by that I mean you just over hire or it's very easy to overhire because you're like yeah I just have money I'll just hire all these people to do all these random things that you may not necessarily need straight away so my kind of main takeaways is that obviously the right timing can be very helpful and as the previous speaker said you got to ride the wave so you know if you're kind of wanting to build a company at the intersection of like Ai and web3 like now is the perfect time the wave is definitely there um another thing is that most VCS don't do their due diligence I've seen that you know in our case I've seen that in case of many other entrepreneurs the friends of mine so they followed the herd mentality as teamer also said and you can use that to your advantage right if you can get one or two VCS that are interested and that can just hype up your idea they can fill up your rounds very very easily and all the other ones will just come out of nowhere all of a sudden saying oh yeah we heard you have a great idea but they haven't really even like looked at the presentation so you know do that do without what you want but so for term sheets again you got to consider carefully um both the current market environment and the future Market environment and that goes back to the safe agreements that I said especially in crypto you know if you're in a bull market now the chances are that in one or two years when you're safe instrument converts you're going to be in a bear Market that's kind of almost like a guaranteed in crypto um so you you got to be careful like what types of um Clauses you put in there and think about okay what if what if the valiation gets cut in like half what happens then and again raise the right amount of money don't raise too much don't raise too little um it's obviously it's kind of hard to know from the beginning what that amount of money is but maybe you can look at some other successful startups and see what they did all right so hiring hiring is a really big one we spend a lot of time doing that and um first off we said okay we're going to be a globally remote company you know we thought it was cool uh we thought it was like yeah I'm not going to be in office it's going to be great it can work from wherever you want but that doesn't really work for a number of reasons primarily because time zone differences so we were based in Europe and we had a couple of people in Asia and basically you know that was already hard enough and then that didn't allow us to hire anyone in Americas because Crossing basically three time zone groups is impossible and you as a Founder are just going to be working 24 7 and that's not sustainable so you got to really pick one or two groups primarily possibly if you can do one you know Americas or Europe um especially for your core team people should really be at least in the same time zone if not in the same country and that also makes things a lot easier if you want to do offsides and similar things like that um you got to pick up employment type now that's uh that's actually uh you know we haven't thought about that initially but then we were like okay you know why don't we just hire people as contractors and that's kind of common in crypto but it's not common in any other industry especially if you want to hire people coming from Big Tech you know they don't want to be contractors they want to be full-time employees with all the benefits and pensions and all that so then you're like okay if I want to hire them in all these different countries I need legal entities everywhere and that's a lot of like legal headache and and issues so then you ask yourself how do you kind of solve that question of legal entities everywhere and there's one thing called eors so eors are employers of record they're basically companies that have companies in all countries around the world and what they can do is they can hire people for you and then lease them to you um even though they're like their employees they're kind of your employees right and then you pay them for that service and you pay them like a monthly fee and all of that so it's really good you can hire people really quickly you can hire them in all countries around the world uh but the problem is it's really hard to fire people once you hire them through that through URS especially if you if you hire them as full-time employees because you have options of both contractors and nfts um so we had a bunch of problems with actually letting people go when once we needed to downsize the team for for different reasons because there's many different labor laws in all these other countries you got to know them all so if you hire in German you got to know German laws if you hire people in France you've got no French laws the eors are not really going to help you they're going to protect their own interest and they're going to tell you oh you know we can't fire them because this law and that law and then you get a kind of negotiate with them basically and tell them no you're the law is actually this or that you know for example we had a case in Germany where we wanted to fire someone because of redundancy you know we simply pivoted and the job that they were doing was not necessary in any way anymore and even the person agreed that like yeah I need to be fired for redundancy so there was no resistance from the employee itself but this this is a valid reason in Germany for firing people but the EUR couldn't take that reason because they themselves cannot fire someone for redundancy because if they do then they can't hire for the same role in the next two years and given that they have like one company where they hire you know hundreds of people for for all the other clients for all the other companies they can't use the reason of redundancy as a reason for firing so you know that's that's a surprise for you and then you gotta kind of plan ahead and figure out what to do around that so that's yours then sourcing how do you find good talent right so that's that's number one thing usually people kind of use their their networks and all of that but in general you know you're not going to find enough people in your networks unless your network is like really huge um so you got to use all kinds of other means you know message boards and and you know finding people hunting people on GitHub and whatever not um using contractors and and sourcers who can do it for you and uh it's a very very long process you know in some cases for certain machine learning researcher positions we had to interview over 100 people and kind of managing that process itself is uh is very tedious so you know kind of setting up a setting up a process using some kind of HR software can really help um and managing expectations is another big one so especially if you want to hire people from kind of traditional Industries like big Tech Industries and so on and which we did for machine learning researchers and Engineers um you know they're not used to kind of being first fully remote that's kind of weird for them because they're used to having an office but you also kind of have to tell them hey you know we're working a bit of a different pace different time zones you know it's going to be different hours and there's no not all these like great benefits that you get at Google and Facebook and so on right so it's more about do you really want to do this kind of job and work in this kind of industry and if not then that's fine right so you just need to explain to them because they ex they have very different expectations so you got to be clear on that um this is a boring one but an important One Employment contracts it sounds very boring but it actually save a lot of hassle or created you a lot of hassle when if you don't do it well so just make sure you have a really really solid Employment contract that you give to people that really defines you know sort of their responsibilities the IP rights um you know all the all the stuff that kind of goes into it because if it ever comes to you have you having to let people go you know that will be your document that you're going to use to basically say Hey you haven't like performed well or you breach this part of your contract or that if you have a very relaxed Employment contract you're going to get sued and then that's going to create a big big problem for your company and the VCS won't want to give you any more money until you resolve lawsuits and so on um so the kind of takeaways from here is um you don't really want to be using URS so you want to open your own legal entity in one or two big markets where you can get the people you need so it's probably America or EU something like that and if you want to use eors for other countries use them very sparingly and use them for contractors because those people can be let go easily the core team should be really in the same same time zone we had the issue where we were split between Asia and Europe and it was it was a nightmare we just had to bring people over to Europe in the end staying lean as long as possible is something I'm going to cover in a bit but essentially it's like don't over hire until you get a product Market fit because if you over hire with a certain idea and then you decide like oh crap I need to Pivot and you really hire like a bunch of people that are doing one certain thing that may not be necessary in your new new idea then you know you have to kind of let people go and just it's never a nice thing to do and the final point is that if you're going to hire from your network just be really careful about considering your own personal relationship with those people right because it you know we tend to I mean I've made this mistake I've hired some very close friends of mine and it's just very difficult to separate like life and personal stuff and it can come and you know this can come in like many many different forms but it can just complicate the business relationship itself so it's always best to kind of hire people who you know as acquaintances but not really a super close friends unless you're really sure that you can work with them uh in a in a good relationship all right um the third thing product so the product itself again a backstory what did we wanted to do as I mentioned we started off as a research company really researching this kind of ml models for automated Market making or news for V3 concentrated liquidity all of that and we spend a lot of time doing that but basically after eight months of research it was clear to us that the problem was mathematically infeasible it's a whole different sort of story why that is and if this is a topic that interests you I'm happy to talk about it later but the gist of it is that the lp problem is essentially just broken and the LPS are not compensated enough for the risk that they take um you know certainly you can hear like oh yeah there's a bunch of people who made money helping and all of that yes they have but in the long term if you want to do this on many many pairs in the long term you're just going to get wrecked right um so we concluded there's no simple solution or any any kind of solution that we wanted to to work on that satisfies the the kind of LP exposure risk so the product idea was dead and we spent eight months doing it so what comes then well then we have to Pivot right so people to super hard right and the chances are you probably are going to have to do them if you're doing a company right like it's it's kind of hard to say that the first idea will always be the idea that that works right so you should really don't want to make them even harder and what I mean by that is that you should stay lean as long as possible preferably until you hit some kind of product Market fit and again cliche but you should really try to build in prod because that will shorten the time that you're building um and then you're testing all of this and it will tell you that your idea sucks maybe earlier and the final thing is that you shouldn't pigeonhole yourself within a certain framework of thinking and what I mean by that in our case it was when we were thinking okay what's the other idea that we want to try what's this kind of new idea we were saying oh yeah we're at the intersection of vetri and AI so the idea must fit within that framework it cannot just be like oh let's just do this more AI stuff that's like not so related to that tree even though we had some good ideas about that we just dismissed them and in my opinion kind of now in hindsight I think that was a mistake um now if you do make all these mistakes that I just mentioned um as we did then you're also in a place where you're pivoting remotely and that's even harder so what do you do then is well you try to get people together on an offside and a hackathon for a couple weeks and that can speed things immensely right that can really get the kind of creative juices flowing and it's nice to get all people together and maybe it's one of the few times that they actually get to kind of interact in person and meet meet each other right so you hack out a few prototypes of different ideas and then later on you can select one that's the kind of most promising idea and the final Point here over communicate with your team this applies can in general especially if you're working remotely but I think in cases where you're pivoting and you know the kind of mission of the company and the vision is changing I think it's really important to kind of be very open and very clear with people so that they don't lose focus and they don't lose kind of hope that this is going anywhere right because it's a it's sort of a trying time um and then the final thing your ideas will be misunderstood and a demo speaks more than a thousand words so we certainly had a number of cases where we had some ideas um that we thought were good and you know we kind of presented them mostly to some VCS which kind of dismissed them very very quickly uh even though again in hindsight now I think there were good ideas we just didn't make a demo to show it to them to for them to understand it because the you know the line of questioning was always kind of going in the wrong direction I think so my main takeaways and some stuff that I will do differently next time here is again look for a problem and then find a solution rather than the other way around we certainly fell into this problem into this kind of mistake where we were saying okay we have this like machine learning technology whether that is you know GPT type of thing or or any kind of machine learning optimization and so let's find a problem in D5 or let's find a problem in in web3 to solve it with right which is just the wrong approach you should find a problem and they say what is the kind of minimum thing needed to solve that problem rather than trying to fit a solution to it again I said always question the framework with between you operate so I've explained that one expecting pivots to happen you should question these assumptions early so you can question assumptions about you know the stuff that I'm doing now for the last like six months in the research is that really you know is that really something that's going to take me to where I want to be um I think Founders in general I found definitely find themselves in this like weird states of Mind where their likes truly believe that this is it you know this is the stuff that I'm doing and this is going to work and you know having a few people around you who are maybe a bit more grounded can always help you um maybe course correct some sometime earlier than than before you fail fully um and always ask other builders for feedback I think that's much better than asking VCS for feedback because feces in my opinion are you know they have some of their own thesis and if whatever you're building doesn't fit into their thesis they'll not be like so creative in helping you get some feedback and then finally uh the most fun thing legal and Ops um I'm going to talk only about one thing here and the one thing is basically the way of doing things the way of structuring things as a crypto company and the big dilemma I've seen when people start companies is should I just go full crypto native so meaning like start a Dao and and you know try to be as as much crypto native as possible or should I just go kind of the standard webto Way open a company you know just do the usual stuff right and I think my answer is somewhere in the middle but let's first look at these two kind of extremes and and see what we can get from that so the first extreme is obviously full web three so you're going full down from from the very beginning um you might have some kind of a holding company that you know can hold some kind of Ip or something but basically people are just hired and paid um you know as contractors they're paid in usdc usually in stables and this allows you a lot more flexibility to kind of move much faster and to scur a lot of bureaucracy that can come with creating a regular company it allows you to issue a token much more easily and you know some would argue that in in in this day and age having a token is kind of interesting and and very very important for competition I mean maybe I don't agree with that fully but certainly we've seen some of our competitors do that and just become a lot more popular and build a community much faster just because they have a token no matter what the product is really but then again you're going to have a lot of real world interfacing problems so it's going to be hard for you to open for example a Google Cloud account right because you know you don't have a company and you're not really going to get us get support from them and whose credit card are you going to use and it's just going to be a lot of hustle there restricted hiring pool again people who come from Big Tech and other Industries they want a normal working environment they don't want to be part of a dial like what is a Dao to them that is weird um and uh and then the final thing is you're open to all kinds of regulatory and legal actions so you know I think there's been a lot of discussion about that in the last year but you know dials are actually can be quite legally liable for for all kinds of things because they don't really have the protection of the LLC or whatever other company structure you choose to use now the other end of it is obviously kind of the standard going full web too you know you have a company you kind of follow all the rules although when you're dealing with crypto you're navigating The Gray Zone a lot there's going to be a lot of places where you're like I don't really know what the rules are you're going to be spending a lot of money on lawyers who are going to tell you a lot of cryptic stuff that you don't understand as well um so yeah that's the problem but it's obviously much harder to issue a token as well it can be done of course and you know many companies have done that but it requires a lot of corporate structuring uh and some kind of layering and basically a lot of dollars to do that but you know it's a lot easier to hire it's a lot easier to interface with the real world work with other companies protect your IP and protect yourself that's basically as a director of the company with the general liability and and all that stuff so my kind of take away from that is really I think this is what I've seen a lot of other kind of successful web3 companies do is that they make a bit of a hybrid here they start with a company and then they try moving towards the Dow towards the foundation uh the foundation then later allows them to issue a token um so all of that is it's pretty complicated I mean if this if something like this actually interests you I'm happy to talk to you about because we've spent a lot of money and time on lawyers talking to them about us of how how you would kind of structure these these types of things but again you know having a token is a double-edged sword and when you want to issue a token the lawyer that you speak to if he's a good crypto lawyer he'll offer you like a whole range of possibilities right you can go with like a really risky gung-ho option or you can go with a really tight legal option and the difference between between that is usually in hundreds of thousands of dollars um so yeah depends and finally yeah there's principles here that are kind of the one size doesn't fit all so you know if you are a project that can actually decentralize quickly and it makes sense for you to be a dial from the very beginning and that's probably the best option right but that's just not the case for like I think 95 plus projects out there so they need to kind of pick this hybrid approach and I think a lot of cases just don't think that because you're kind of starting a Veterinary company or or even a Dao none of the kind of classic business practices apply to you so you still need to do all kinds of uh you know accounting and and all of that stuff that that goes with it because if you have any kind of easy investors they will ask you for your quarterly statements they will ask you for like all kinds of balance sheets assets and all of that um so you got to do that and uh yeah final point is hire a good crypto lawyer I mean there are far and few in between but some of them can really help you um but again don't always listen to them because they are lawyers in the end no this disrespect to any lawyers here if there are any but uh you know they always have their own sort of point of view and and they'll just say like yeah this is kind of Gray Zone this is legal this is maybe not legal and you're just going to kind of pick up yourself and decide what uh what is the amount of risk they want to take and that's all I think this is just the quote that I wanted to leave you with so always take everyone's other entrepreneurs lessons with a grain of salt because everyone's path is different and the problems that you face are going to be probably very different all right any any questions [Applause] uh thank you thank you for this speech very interesting very useful even though it's not about web free or AI but about how to run your own company but it's it's amazing and I'm like happy to see you had this journey and experience and the question is about that right amount of money that's you need to raise and even though you say that it's really hard to understand like what is the right amount but maybe you can share about some you know criteria maybe you know there should be money to spend for your core team for 20 people for one two three years and it will be enough for any other ideas for criteria thank you yeah I mean that's super good question um there's no single I think set of criteria here but something that I think about or that I would think about next time is you know what is the minimum number of people that you need to build a product that you're building um that's definitely not 20 or even 15 or 10 like I think you can build with five like five good Engineers is what you need but you need to find those like really really good Engineers those 10x engineers and they're going to cost you right so you got to figure out okay how much are they going to cost me you want to kind of have a runway for two two and a half years um two years is the kind of usual recommended but you know now if you're in a bear Market in crypto you might want to extend that to three um so calculating that getting the salaries getting the number of people they need I'm thinking about whether you're going to need a lot of compute so if you're building a machine learning startup you're going to be spending a lot of money on that and then you can get to a number that is uh sort of realistic I would guess I mean and then of course add on top of that any kind of unpredictable um things that happen amount and obviously the legal amount because in crypto that's going to cost you a lot especially if you're wanting to do this corporate structuring around token you know you have to do things that will that will run into six figures so in that sense I would say you know three million is probably enough to to kind of Kickstart a company because you know pivoting is is really hard and and that is one part where I think it's better to just run out of money and try again rather than kind of try to you know pull the whole company in a totally different direction and kind of restructure everything and it just it just brings a lot of difficulties and if you just get enough money to build the thing you're planning to build with obviously some leeway to to get some kind of small pivots done that should be enough right and then you just run out of money or you succeed it doesn't matter you can try again right rather than saying oh no now you know we have like still a million and a half left or two million left we can try something new but you already spent your 3 million and you basically owe those three million to your investors right so now even if you if you make success with the other two million you've kind of kind of gambled away those those you know 60 of your investment on something that is not useful um so that can kind of come and haunt you down the line I don't know if I'm making any sense there but um you essentially want to use the money that you get as much as possible for the thing you eventually end up building and for a thing that makes you money rather than on kind of all the other side Explorations that will not really net you money so if you don't have a lot of it you can then you just really focus on the thing that is important uh hi thank you for the nice presentation uh I have question uh like uh I assume that you will begin a new company and something like that and what are the core values that you will look into your next partner for that business and or what are the questions you will ask your partner before going into partnership I love that question that's that's a difficult one um yeah I think the problem this time is I mean I I love my partner he was he was really really good um but the problem is like me and him were very similar uh and that's that's the kind of probably the biggest issue right we're both Engineers we were both worked at deepmind and we were both really good at AI um you know I certainly had more Affinity towards the kind of business side of things and and all of that but it was not enough like I felt if I had someone who was actually really good at like marketing and community and and kind of Business Development would have helped me a lot more um because I could have handled the technical part right um again if you have some a technical co-founder then you can really think about you know you can kind of bounce around a lot of ideas on technical size and inside and that can be helpful but I would say yeah just look for complementary skills and uh you know I can certainly be sometimes a bit less grounded uh in a sense and and just a bit hesitant to kind of make a decision so whoever can kind of complement that and whoever is like more quick on the on the trigger let's say I think that's that's a good combo because if you have two people who are like that then nothing gets done in the end you know and like what are the red flags you would like immediately let's say pass on the opportunity let's say you mean yeah yeah for the yeah on the co-founder I think there's a lot I mean there's a lot of things that you just see um but I think it's basically people who who really crack Under Pressure really quickly and have no patience right um and so that's one and actually maybe even more important one is someone who cannot have a discussion um and a debate right because that's really going to be like 50 of your time when you're discussing with your co-founder you're not going to agree on something um you know and someone who can take the feedback and say okay fine um I can see your point but maybe I don't agree with this but you know let's let's agree on a direction right that's super important if if you can't get to that then just the company is not going to go anywhere right um and maybe maybe another advice I think is don't do 50 50 Equity split I think that was sort of a mistake that we made because then there's no leader like you you have no no one to say to just cut and say this is how we're going to do it um so I would say at least 60 40 or whatever kind of makes sense but someone has to basically take the reins and say I'm I'm running this company and and you know my word is the last one um not saying that you should be a dictator right but but kind of also yes sometimes and it's really amazing I just wanted to say again that it's really amazing that you so highly introspective and I'm guessing it comes from overthinking because I recognize it but um I mean the amount of lessons that you extracted for from from these experiences trust me it takes a bunch of entrepreneurs a lifetime and they still don't get it so it's it's great that you put on you know you know what you're good at you know what you're missing and you're gonna get better and better a lot faster it's it thank you very much for sharing all of this with us it's a curse and a gift as everything is foreign thank you great presentation I have a question like you selected to build a company based on the AI which you were very proficient like an expert and web3 so what do you think about web3 say if you started with AI it's a manufacturing well it has bigger chance to succeed or not like the selection of web 3 was it was it the right step or it just confused you some somehow very good question um I mean just gonna I gotta preface it with the fact that you know I came to web3 from my ideological side I've been in crypto since like 2015 and you know it was more from uh from the kind of Anarchy side that I came into it and it's been part of my life ever since I've kind of even worked in Ai and deepmind so my wish was really to bring these two things together um maybe that's not the best sort of way of coming into a business from the ideological side but you know I wanted to give it a try and I think I would say especially a couple of years ago when we did that there was just not many good use cases for AI and crypto and we've definitely find that out during the during the kind of a lifetime of the company nowadays with GPT three and four you're definitely seeing people saying okay you know we can use it to kind of help people build smart contracts code do some search that's all good those are actually like good use cases zkml can also be good but although I think it's still early days on that um but you know before anything that you wanted to do with with ML and and crypto had to be centralized nothing still is right because how are you gonna you know you're gonna build some fancy model on some data and then you can plug it into some kind of a crypto service it's it's just not going to be decentralized right the only way you can make make this work is in cases of like Mev block building trading you know any of those kind of centralized type of things that you want to do with AI I think to get to the point where web3 and AI are actually meaningfully working together in a decentralized way I think it's going to take us some years still I don't know if that answers the question uh hi here in the back hello hi uh great presentation great talk um in one of your last slides you had this uh sentence that really resonated with me uh it said something along the lines that even though you're working in web 3 you know don't think that the rules of you know previous businesses don't apply here I would like to play The Devil's Advocate and ask you for your opinion on if you think that the web 3 Industry as such is not as successful because we are not applying enough lessons from the hundreds of years of Entrepreneurship that you know came before that we are you know mystifying and pulling you know like all of these things around web3 into like oh this is some thing quite esoteric but at the end of the day you know it's very simple money must come in money must go out costs have to be you know uh paid and so on like what's your thought on that yeah I'm not sure that's the Devil's Advocate really that's uh that's pretty much what I agree with um yeah there's no um I mean the unit economics and just the general economics applies to the spaces wall right like I think people got this idea that we can just print money out of nowhere because we print tokens and all of a sudden they have value and you know where does this value come from it comes from people who are just want to speculate right um so I think speculators have really driven the economics of crypto but if you take if you take that away as as you do in a bear Market you just see that nowadays races are asking you for like oh do you have traction do you have users do you have Revenue you know in the in the bull market nobody asked you that they were like oh yeah you're going to issue a token great here's money right um and and it's just a bit of a broken model right like there's no I'm not saying of course there's companies who are doing that but the problem is there's just not enough users in web 3 and if you don't have users you're not going to really be making Revenue I mean the the few protocols that are actually making Revenue like I don't know uni Swap and curve and and so on like these taxes they're catering to the main User Group in crypto which is gamblers right and speculators um I think everything else that we've seen has been sort of ethereal you know things like gaming and axi and all of that again it's been people who are not really users in the using the game because they like the game they use the game because they can earn money so it's sort of a subset of that I think for crypto to go mainstream and to actually get users and to actually obey the economics and not just the boom and bus cycles that come with VC money going in and out uh it will require some kind of actual useful use cases and that's just about it