Will AI Agents Use Crypto? | Qiao Wang

Transcript

so fundamentally AI agents are are programmable so they can leverage other interfaces to do things that they want so one of the interfaces that you can use is the blockchain and they can do this permissionlessly so you so for example they can use eth as the AI native money or they can interact with D5 or they can play games and if they do that they can do that permissionlessly and the one can stop them so that can get a little bit scary this episode is brought to you by carbon carbon is a new Dex on ethereum that makes concentrated liquidity super easy with carbon LPS can now automate your liquidity strategy with custom on-chain limit and range orders all from a beautiful UI check out

carbon today for unprecedented control over your liquidity please all right everyone welcome back to another episode of Empire uh very lucky to be joined by Chao Wang from Alliance today chat with American strawman thanks for having me yeah of course of course um one of my favorite you're one of my favorite people to talk to about uh broader just how you're feeling about the market kind of state of the market um and what you're seeing really interesting and unique background from from Alliance I think they're kind of like three places that we could go in this conversation there's almost like state of the market if you know we're 18 20 months into this bear market now would love to maybe like just kick it

off with how you're feeling about things there's also a conversation around like I think you and some of the team have been leading the conversation around Ai and crypto and the intersection those two kind of things converging and feels like and then there's a third bucket which is you have this really unique experience from Alliance seeing what chains that teams are building on and what teams are building and why they're having success and why a lot of them maybe aren't having success so I'd like to go there as well but let's kick it off with high level um 18 19 20 months into the bear Market how are you feeling about the market today um so there's uh several aspects we can talk about one is um on the uh Venture side of things on the private private primary Market it's

basically dead um I don't if if you saw the uh chart from I think it was um difalama about the month by month uh Venture funding in crypto I think we're at the lowest level since uh the last bear Market and uh in terms of dollar amount um and I don't even know if it's the bottom um and you know no one is in a rush to to uh to write checks right now uh so valuations are going down still um and fundamentally it's just the fact that the private Market generally lacks the the public market by maybe two three four quarters so even if Bitcoin has gone up 2x this

year uh the private Market has been bleeding actually slowly uh so so that's what I'm seeing in in the primary Market the founders are having a very hard time in fundraising um Public Market I really don't know I mean today trade Bitcoin is trading at three thousand thirty thousand and it feels a little bit heavy because we we couldn't we couldn't pump on good news so we had like really good you know Bitcoin ETF news uh we had the xrp news uh which was a huge victory um and um somehow we can go above 30k uh so it's a little bit worrisome and then on the macro side I have no idea you know half a year ago people

were really bearish you know it was the most anticipated recession of all time but then today people are like oh maybe we're gonna get the soft Landing I have no idea I don't have an edge in that market so um we'll see yeah tell me more about the private markets in crypto it strikes me that there's actually funds who might have some cash and they're sitting on some dollars either they are sitting on cash and don't want to deploy it or they're almost timid to make a capital call to their LPS and it seems like that's the reason that companies aren't getting funded is um they actually have the money they just they they just won't make the commitment and make the check but tell me if I'm right or wrong there

okay so there there's a few possible reasons I don't know for a fact but the few possible reasons are number one um they do have the cash but um a year ago they yoloed into a bunch of you know uh deals at ridiculous valuations um and now they just can't explain to their LPS um they have to do a lot more due diligence also a lot of them invested in in some ridiculous failures catastrophic catastrophic failures like FTX for example so now they're doing a lot more diligence due diligence and they're deploying a lot slower but they do have the money they're just not in a rush to

to deploy um another possible reason is that um again back to the earlier uh point about the private Market lagging the public market so they're still feeling it out so they they are everyone is aware that evaluations are continued to go down but no one knows when it'll bottom um so the one wants to catch a falling knife a very illiquid falling knife um so that's sort of another reason but it personally it's really mind-boggling that you would expect the vast majority of VCS to be uh the kind of you know contrarian investors you know be greedy when others

are fearful kind of investors but that is not the case right now everyone is fearful um half a year ago when FTX happened I thought that was the bottom because I've never seen anything like that in the history of crypto like that was the worst in my opinion the worst event that has ever happened to to crypto I thought that was the bottom for the pro for the private Market but it was not like things are continuing to go down yeah it's interesting I think you're seeing a real Divergence in the private markets in crypto right now between the series the seed and series a companies and the um you know the series B series C and series D and the series C and series D are getting whacked right valuations are cut in some instances 90 right maybe they raised it 10 billion now they're raising a little chunk of

change at you know one 1.5 right they've been completely whacked I'm actually I don't know if you're seeing this I'm seeing very interesting seed deals right now um and I'm seeing seed uh seed deals in fundraises get done right I'm you know maybe then maybe it's a new L2 or something or maybe it's a consumer app and it's a it's a reasonable evaluation it's 10 15 million dollars and I think that's they can just get funded purely with maybe a couple very crypto native funds putting in a million or two each and then a lot of angels and I think there's this those deals are happening but the later the series B the CRC are completely dead right now yep that I think that's exactly right and the 10 to 50 million range that you mentioned is spot on us what I'm seeing as well on

the seat at the C stage preset is probably a lot lower yeah that's where we are at for example we prefer it precede um but uh and you're right that um the later stage deals are not happening um because everyone got wrecked all the latest hvcs got wrecked that that invested at um at the peak of the bull market um but in general deals are there's fewer deals um I also think there I also feel there's fewer interesting projects and that might be a result of just fewer developer activities uh in the whole crypto ecosystem so I don't know if you saw the uh the report from Electric Capital

um so they have a live uh data set on the number of new developers going to crypto and the number of active developers in crypto and those two numbers have been down 25 to 50 respectively um from the peak which was uh about a year ago so there's fewer people building crypto right now a lot of them are pivoting to uh to AI for example yeah which we'll talk about I'm I'm not gonna force you to predict uh the macro markets here chat but I'm gonna force you to give me a little prediction on on crypto markets yeah um we've we pulled up uh if you can see can you see this right here

yep so Mike and I shared this we've do a company offset at block works every six months um we shared this at the July 2022 offsite we said bear markets come in four phases phase one is the the blow up right this is the complete unwind Euphoria rolls over leverage blows out there are these rapid price changes down and the enemy in this market is fear everybody is just terrified because things are collapsing so quickly stage two which is where we were at in July 2022 we argued was the Fallout or maybe otherwise that it was forced capitulation everyone is angry people were looking for someone to blame you know this is when a lot of angry tweets about dokwon uh machinski right

and the enemy here is disenchantment just like what are we what are we doing here just this is you know this is a real bummer right stage three which I would argue we've been in for the last several months is boredom this is bottomless exhaustion complete silence lack of attention things bullish news happens but nobody cares Peak floor and attention and the enemy here is is boredom um I would argue that we are now coming out of stage three and into stage four which Mike and I have kind of deemed the rebirth stage this is when Sparks of Hope get brushed off this is when Ripple beats the SEC BlackRock announces a Bitcoin ETF but because the enemy in

this stage is disbelief it doesn't do anything to the market and I'm curious where are you if first off if you agree with these stages and then the second more important question is do you think that we're kind of heading into this stage four rebirth stage right now yeah so um I understand you're a stage one two three I don't fully understand stage four it uh by the way I agree with your one two three uh but in my mind stage four to me it's the same stage as as through so I I can't I can't make it this distinct online but I feel like we're in the for me we're still in the apathy boredom phase uh and you mentioned that

that the good news get uh brushed off uh which is which sounds like apathy to me um and um you know I feel that because I whenever I tweet these days I get very little engagement I feel like I'm yelling in the dark no one listens to me um so uh I feel like we're at that stage so that we're still in the so that stage is where the market bottoms it's a very long and um arduous um lengthy uh bottoming process so that usually takes Maybe let's say a year if not more uh add this according to the to the last couple of Cycles but the thing that makes this cycle really

different is um the macro like where the whole crypto ecosystem is a lot bigger than the last cycle so there's a lot more wealth effect uh that leads to Greater correlation between crypto and macro so I can't I have a hard time predicting where the public market will go in the next six to 12 months I think a lot of it has to do with uh uh with the macro earlier this year at the beginning of the year I I tweeted that um I think will be a lot higher at the end of the year than at the beginning of the year and so far I'm right um but from this point on I really don't know I think it's really 50 50.

yeah I think one of the interesting things to watch next year is the the proverbial fight between the macro especially if we have a bad macro environment next year this battle between the bad macro and the Bitcoin having right because historically crypto has obviously only existed in a pretty good macro environment and every four years we have this Bitcoin having which the four-year Bull and Bear Market Cycles are very tied into this having I remember exactly where I was in the 20 in May 2020 having during covet I remember the seat I was sitting in I remember what it did to the market I remember what was happening on Twitter that day and I'm looking ahead to this April maybe May 2024 Bitcoin having

which could very well kick off this next Bull Run however you've got this this macro that nobody has any idea what's happening so I think that's gonna be a very interesting kind of convergence to watch there yeah it's so tour plan is entirely possible that um we've seen the end of the regular four-year Cycles it's entirely possible that this upcoming cycle will be either shorter or longer than the usual four years foreign against the four-year Cycles yet I don't think we're there yet um John maybe if you could give us some insight into what folks are building right now that'd be I would I would love to hear this so you guys have gone

through um you guys are on all 11 I think it is the 11th cohort um yeah so big congrats on that I remember when you guys were doing doing the first one yeah um what if what are you guys seeing right now in terms of what folks are building in in all 11. yep um so things that you hear on Twitter is mostly right that there's a lot more infrastructure projects than um consumer facing products um again this is very cyclical and it's historical this has to be the case as well that a German bear markets people build more infrastructure than consumer products um and so currently by infrastructure I

mean for example again scaling right um you know restaking um a share sequencer like that kind of stuff right especially for the ethereum ecosystem but there's also a lot of infrastructure around wallets um and account abstraction is a Big Driver of that in general I feel a little bit um uninspired by the startups on the consumer side I think there's just fewer of them um and I love to see more and I love to see crazy ideas ideas that that piss a lot of people off um ideas that are really creative that

are potentially even outrageous like social networks that Embrace Bots instead of pushing Bots away yeah for example well by the way um so um Nikita beer this so he's uh one of the Legends in the consumer products in web 2 and he tweeted something really interesting today which is that um it's about AI but I think it's also transferable to crypto but what he said was that um the AI consumer product startups are building something most of them are building things that are too obvious so for example AI girlfriends but uh he also said that um the truly good Founders are built are

using AI to bootstrap the the critical mass for their dating app if that makes sense so you know if you build a dating app initially you have a huge uh cool star problem right there's there's you need to supply and and the demand basically the men and women so how do people strap that co-star problem you use AI to do it use boss to do it um so um Yeah just something really interesting that I read today I think that's right by the way Nikita is about as smart as they get when it comes to building these consumer social apps right I think you start gas app sold to Discord uh what was it he sold another one like the B app or something it was I think it's the same app and he built it uh within the

span of like over five years so like five years ago they built the first version two sold one to Facebook sold one to Discord same app yeah yeah exactly so all right so um so folks are building infrastructure so I so I had a feeling that you're gonna say that what um when they're building infrastructure are they building is this like more l2s is this a kind of abstraction is this ZK is this what I'd almost call Deep infrastructure Tech um that is like very technical is this more like hey we give you a seamless way to embed account abstraction into your wallet like what don't tell me let's double click on the infrastructure all these things that you mentioned a lot of ZK

um a lot of scaling Layer Two oh you know a cello rebranded or not rebranded but pivoted from being a leader one to Layer Two um last week no I now followed that very closely I think I think that is first of many l ones actually that realize that managing your own nodes and validators is just a complete pain in the ass and that it's uh that it's just easier to share the security of ethereum and that yeah I think that I think that will continue happening for some of these other ones yeah what did you say what do you think of that chair um I think it's uh from from an outsider's point of view it's a very intuitive

um pivot it makes a lot of sense um ethereum is where the users are and you want to capture those users and the only way to do that is um if you don't already have a lot of users like Solana does uh it's really obvious to Pivot into a ethereum layer too infrastructure there's a lot of deep tag in ZK uh scaling um and scaling related but there's also wallet infrastructure around uh account abstraction I don't think that's super deep Tech um it's often uh Like Rappers around existing the attack um and and making it very easy for developers to build on top of um there's zkml as well there's uh a

niche of developers building um products at the inner or deep Tech at the intersection of zknai for example um but yeah that's what I'm seeing yeah what um I remember after I don't know if you guys still do this but after recent cohorts I think it was like six and seven and eight you would send us you would send uh investors a message and you're like these are the these are where folks are building these These are what chains people are building on these are what you know where they're building in the world I don't know if you guys hear that data publicly still but I'm very curious like are folks building on l1's are they mostly building on l2s like how are they making those choices maybe as much data as you guys have on that I'd be very curious to hear it so

um shortly after uh FTX last year um Solana went from number three to number four um so Solana has always been number three uh right behind polygon and uh the ethereum mainnet um and then when it went from number three to number four it was arbitrum that took Salon its place so that was half a year ago and then shortly after um people recovered the Solana ecosystem recovered so they went back to being number three um overtaking arbitrum but recently they're back to number four again uh

overtaken by Arboretum again so it's a battle between salon and Arbors I don't know how statistically significant the that ranking is but um I think the overall ranking hasn't changed the the big five so is the big fives are ethereum Menard uh polygon Solana arbitrum optimism and they're far ahead of everyone else hmm that lines up with how I would expect the market to be why do folks build on ethereum Main net instead of building on arbitrim or optimism um I don't know to be honest yeah I don't know yeah

well are any what about l3s like are any is anyone looking at what dydx is doing with Cosmos and saying maybe this is a better strategy for us maybe maybe we do need to know to own the full stack I I haven't seen anyone that adopted the same uh strategy as cosmos yeah and um I don't know what to make of that that's the dydx strategy it's what we'll see so they're gonna launch I don't know where they're gonna launch their their next version that's fully Cosmos build that's the V4 I think it might be in a couple months October October okay um but uh from the looks of it you can use dydx from any ethereum based uh wallets including metamask so so I would

imagine that the user experience will be very similar uh as if Cosmos was on ethereum oh sorry as if dydx was on ethereum yeah yeah because originally my biggest uh worry was that they moved to Cosmos but they now have access to a much smaller user base but from the ux it doesn't seem that way so I don't think they're cutting off the liquidity though to the to the east the East liquidity basically so yeah what um what what uh what types of Founders are coming in is it I don't know if you saw Rob leschner launch his new company super State um yeah that was a very my mind encouraging development because you have this like very uh very successful crypto founder we're now

at the stage of the market or crypto's been a lot been around long enough or you can have successful crypto Founders launching their second thing which if you look at back at you know Tech over the last 20 years a lot of the best companies started from Founders who are not first-time Founders their second time or even third time Founders in Tech um so it's very encouraging in my mind to see the lessons of the world do this I'm curious what types of Founders are coming into it to to this cohort in the last couple of cohorts is it web2 is it crypto Founders on their second run who is it there's always a smart small percentage of very successful Founders um uh in every cohort so the next one there will be a Founder that previously built a unicorn

in web 2. the last cohort there was uh someone who built a very successful crypto uh network uh that was um David voric from who who built um Sayo um Thai used to be like top 10 coin on coin market cap so every core there is a small percentage of that um but the more interesting trend is uh the next cohort virtually all of them are crypto natives uh not surprisingly compared to two years ago at the peak of the bull market there was a lot of uh Fang Engineers or what two people um many of which have pivoted or folded um from from crypto and I feel really

good uh today because at the peak of the bull market I I always uh worried about whether or not that the founder will be committed to crypto yeah when the bear Market hits um but today I'm not worried anymore because everyone is a crypto native and has been building for a long time in crypto yeah it's been interesting to see people get wiped out I mean I remember you had this tweet at the beginning of the year you said for crypto professionals the second year the bear Market tends to be the most painful one it's the year of apathy there seems to be no light at the end of the tunnel but simply being aware of this gives you a massive Edge year three things will get much better and I've definitely seen that with companies that we that I've invested in where there's this

Divergence of folks who maybe they've been in crypto for a while they understand that this is just you got to grit it out and then the new new folks are I mean a couple of them have pivoted into AI I'm like guys what are we doing here so yeah exactly I'm seeing a a far greater number of uh pivots from crypto in the last few months I've seen more pivots from crypto in the last few months than I've ever seen yeah in the last four years and I've also seen more co-founder breakups in the last few months same I've ever seen same the bear Market is brutal very tough yeah um what about uh huh yeah that's really interesting why are what does the dispute that Founders

have do you think um okay this is a really good question and I think not enough investors pay attention to this um you know I pay a lot of attention to the co-founder relationship um of the co-founders I ideally there are a couple of people who have worked together for a long time in the past um you know within the same company or uh worked on like site projects together and having friends so that would be an ideal setup of co-founders because if they found each other uh online you know some sometime in the last six months um they're still in the

honeymoon period and it's their their relationship will get tested in the bear Market when things don't go well and um oftentimes is really because um the product doesn't grow and one of them just um either loses conviction in crypto or conflicts arise between the co-founders growth solves every problem and conversely the lack of growth causes all sorts of co-founder conflicts it's re I it's something that I never thought about when investing is the co-founder relationship because Mike

Mike and I have a very strong relationship and you know we went to we went to college together we were roommates in New York for a while we'd known each other for several years we handle I think when things do not go when we we disagree all the time obviously as co-founders do but I think we have a very good system of working out disagreements and we move through disagreements very very quickly but so I'd never I'd never thought about like basically interviewing or assessing co-founded relationships until now we're deep into this Market I'm like oh man that's one of the main indicators I would say for how well a company ends up doing yep agreed yeah um let's talk about tensor for a second so I'm at the tensor Founders for the first time um two weeks ago maybe

obviously why uh online we were just uh we've been exchanging DMS for a while on Twitter and we ended up jumping on a call and I'm a big fan of what they've built and it's remarkably impressive how quickly they build and I just the market share that they've that they've grabbed basically in the nft space in in such a short amount of time with such a small team I'm curious when you see a team like do you have you done this long enough now ciao that you can see someone like that you can see that team and you're like that's that's a winning team they are going to do it or is this something that you have no idea it's a complete you know rant random uh random instance here where like that you know just one of your 50 companies ends up

being very successful there yeah um so after doing this for three years I think I I've developed um some Instinct for um people like the tensor guys so matter of fact we actually have a name for those people internally and we call them tensorians um so we're always looking for Founders that are to have the the same characteristics as um as the the two co-founders of tensor um but it doesn't mean that we're not open to other type types of Founders but the archetypical founders we look for these days are uh you know they're fair to Green uh they they're usually first-time Founders because that's usually where we can make

the most impact um they think about problems deeply and they work extremely extremely hard so the tensor guys I don't think they sleep uh and they work seven days a week truly they I was like you know I was asking are you coming down to New York they're like why would we leave I think they're in Toronto or something or Monster they're like why would we leave here we've got our computer we've got our keyboard why would we that that's all you need I was like all right exactly yeah and um but but the the hard word part is very hard to um gauge just from an interview um you have to look at their past achievement yeah but the thing that you can look for during the interview is how they think about the product how they

think about go to market do they have very unique insights about their users that um that I've never heard elsewhere um again like um it sounds simple on on paper but it's most of it is really really is a black box in my mind um it's um it's like a neural network that I develop um from the last just doing this for for three years what are the characteristics if you're sitting down with Imran and Jacob and the team and you've got to whiteboard up and you write down five characteristics of tensorians what's coming up there uh the single most important one is resilience and that it basically trumps everything else

um I think resilience is is the only common uh characteristic of every successful founder every founder is different but resilience is the one characteristic that that is common to everyone so that's the one that we look for but it's also the one that's the hardest to gauge from an interview again yeah um but um so that would be number one number two would be um unique insights into their users so oftentimes uh you hear about a super interesting narrative on Twitter uh it sounds great um but you have to dive deeper you have to understand you have to see if the founders truly understand their users

um you have to go well below the surface so unique insights into their users is um is really uh important so to give you a concrete example the the unique Insight that tensor had about their users was that um the vast majority of trading volume of nfts came from speculators so people who treated enough to use as strict ones not the the collectors the collectors don't trade that much it's the it's the degens that trade a lot a lot in terms of number of transactions but also in dollar amount and you had to build a product that um served these people really well so that was the unique Insight that tensor had

by the way it's the same inside as blur it's exactly the same anything else on tensor no go ahead I'm just gonna say any other successful companies that have come out of Alliance that you've like maybe gained a bit of insight into how you guys can make better Investments moving forward um so there were um here's another really interesting example it's step and steppen came out of our program I think the cohort before tensor and um what makes Stephen really interesting is their English is completely broken

and they would have a very hard time raising from um American VCS um I I chatted with them in Chinese so the the cultural barrier and and the uh and a language barrier is is real um and I think it's an alpha that that I have just because I I so English is actually my third language um before English I as I mostly spoke French in my teenage years in Chinese in first 10 years I grew up in Montreal so Montreal is uh bilingual um but mostly French Ash never spoke I didn't speak English

on a regular basis until the age of 18. but anyway so what I'm trying to say is there's an alpha about um understanding this cultural barrier and break that barrier so um but the Stefan guys have they're this they have the same characteristics as tensor in terms of resilience so these guys also never sleep they're based in Australia they join our program our you know our lecturers at like 1am their time um weekend week out uh it's ridiculous to be honest um so so resilience yeah yeah resilience and hard work um one one more question on the on the alliance cohorts and then we can move on

to some of this AI conversation that I know we wanted to have um it feels like we might be at Peak Bowl Asia bear U.S in crypto right now um I'm curious what you what you're seeing in you know on the ground in terms of like where folks are building where people are trying to get users it feels like there's a lot of focus now on maybe Korea or Japan or renewed interest in Singapore Hong Kong is opening back up um everyone's very anti-us the SEC is clamping down cftc Etc in my mind it feels like we're kind of like I would be betting on the US right now

um but I'm very curious what but that might just be my like U.S Centric group in the U.S world view that I want to have I'm curious uh both like what you think about that and then where like maybe on the data like from the data like where people are actually building so I I certainly thought this way uh a couple of months ago that we were at the you know where Peak Asia and bottom us but I felt like the Bitcoin ETF news and the xrp News changed uh things a lot um so the very fact that it was later think the the BlackRock that filed for the Bitcoin ETF um uh probably would change the perception

of a lot of people in the US from chat fight people to tech people uh that was a huge positive for for the US and then same thing goes with the xrp news uh it was a huge victory obviously the does hasn't fully settled we'll see if the SEC will uh appeal the decision but just from a perception point of view I think the US is actually uh going back up um and then but in terms of uh data I don't have very good data I don't have like clear data I don't know who um you know whether or not Asia is rising in terms of developer activities versus

the US um but I would say uh a good portion of our applicants come from Asia like maybe a third of them were still mostly U.S like most of our applicants are still based in the US like I would say 50 of them are based in the US and then maybe 20 based in uh Europe and other regions oh the one interesting thing there's two interesting trends that that uh emerged recently one is Latin America I think Latin America in terms of crypto adoption is Maybe a few years behind I don't know how much exactly but there is a a far a far greater number of Latin

American Founders building what the U.S Founders were building a few years ago for Latin America so for example on ramp offram Dubai is an another interesting one there's a lot of Founders based in Dubai um some people say there's a lot of scammers in Dubai as well just I suppose there's a reason for that um the the maybe that the taxes and the you know other uh uh other laws but uh there there's more and more Founders is in Dubai and there's certainly more and more founder space in Hong Kong hey everyone we'll get back to Empire in just a minute but before we do that I

want to let you know that we have permissionless coming up permissionless is Big conference that block works and Bank lists put on together it is the biggest the best defy conference in crypto this year it is in Austin Texas September 11th through 13th if you've been in crypto for a while you know that bear Market conferences are the best kind of conferences we have a phenomenal lineup of speakers a lot of the guests that you hear on Empire are both going to be speaking there you will have the opportunity to meet them there and a lot of the topics that we cover on Empire ZK attack roll ups account abstraction Mev app chain thesis a lot of that kind of stuff that will all be discussed at

permissionless this year so because you are a listener of Empire you get a special discount that's right Santi and I negotiated with our marketing team you get 30 off if you go to blockworks.com forward slash permissionless Empire 30 is going to get you 30 off your ticket today when I'm recording this that's about 300 off your ticket so type in Empire 30 when buying your permissionless ticket you get about 300 bucks off click the link at the bottom of this episode it's in the show notes do it quickly because prices go up all the time and if you are going to permissionless hit me up let me know shoot me a DM on Twitter I would love to meet up with you there let's face it concentrated liquidity is hard and that's why I'm super excited to partner

with carbon for Empire carbon is a new Dex on ethereum that makes concentrated liquidity easy with carbon LPS can now automate your liquidity strategy with custom on-chain limit orders and range orders wanna buy a token when it dips and sell it when it spikes with carbon you can now set a strategy that buys in on one price range and sells in a higher range on repeat using a single source of automated rotating liquidity strategies can be created for any standard erc20 token I recently checked out the carbon beta that just dropped pretty Blown Away by the liquidity strategies that carbon enables on chain it has these rich trading features that you'd expect from

a centralized exchange except carbon is only on chain decentralized and non-custodial just connect your wallet it's carbondefi.xyz that's carbon D5 dot XYZ choose a trading pair set your buy and sell ranges and amounts hit create and you're done carbon automatically moves your liquidity into your selected ranges as the market moves LPS it is time to take back control of your liquidity with carbon check out the link and get started today now let's get back to Empire ciao let's move away from Alliance and let's talk about uh let's talk about AI for a little bit I think um I got a little smile on my face saying yeah you got a little smile I

think one of the reasons we just in like uh subconsciously smiled there is it feels like this is kind of like hypey thing and even like bringing up like oh this convergence of AI and crypto it feels very I don't know just fluffy I would say maybe if you could tell me like you know the Gartner hype cycle you know big up down slowly comes back up where do you think that we are on the on the Gartner hype cycle for AI right now I actually tweeted about this uh probably two months ago uh it felt like we're at the peak but it I did caveat that um the bottom my last only a few months because AI is moving really really fast um so today so two months after my tweet

it feels like a lot of people are disillusioned with uh um with uh the current state of AI especially the consumers if you look at the the Google Trends um the uh search for chat apt has been going down for the last couple of months um and then there's also a very very interesting study on how the performance of chat GPT has gone down in the last few months um I thought that was really interesting um but anyway that's that's a different topic um so performance it feels like performance of tragedy like it's how good it is at actually doing its tasks is getting worse right so how good it is in terms of doing math in terms of like just doing

some basic um uh tasks um went down a lot over the last few months and we're actually discussing this internally last night uh about the hypo like what are the hypothesis as to why this this has been happening so one of the hypotheses that we've seen we saw on Twitter is that so openai uses this thing called the um uh reinforcement learning with human feedback so the human feedback is the important is the keyword here so when opening AI first launched uh there was all the human feedbacks they they had to train their model was their internal staff but once they've launched um the product there's like hundreds of millions of

people uh providing this feedback back to open Ai and those people are I don't know they destroy the model somehow I don't really buy this hypothesis but I can see it but another more I think more plausible hypothesis is that openai is prioritizing AI safety versus performance so they actively dumb down the uh the model itself just so people don't do crazy with a very powerful tool so that's another plausible um hypothesis obviously the opinion eye people have denied these things but anyway so that's uh it's just overall this degradation of performance

um these people to feel a little bit more disillusioned with AI in the last couple of months so I feel like there has been a little bit of a sentiment change but that's just my perception I don't know if um other people see the same thing yeah I I think you're right there um so just for background from our past conversations I think like one of the reasons that maybe you're qualified to talk about this is you I mean you have a decade of experience building some of these models I don't think you've like maybe it's not necessarily neural networks because you were mainly working in like building model trade trading models and finance where neural networks don't either they don't really work or they're just too hard but you know you have about a decade of building models

I'm curious how you think about um like when I think about the you know when I see like headlines about the intersection of AI and crypto and when I try to think about it I I've kind of bucketed it into three buckets in my mind there's AI that ends up helping with the AI that helps companies and people inside of companies build better products maybe this is AI that helps developers move 10x faster this is AI that helps customer support chat bots on the website that's like that's one bucket in my mind then there's a more probably two other like maybe more technical buckets which is kind of decentralizing AI and looking at that from from two different angles there's the um uh there's like the compute like decentralizing compute I know the

multi-coin guy is invested in a company called render been looking into them like there's decentralized compute uh there's uh so I guess maybe there's another bucket there's there's training right there's like the how do you decentralize the training of the AI and then there's how do you decentralize like the whatever the front end is on the AI inference the inference yeah thank you there's like decentralizing that so I don't know like how do you what do you think about those buckets is that are you do you agree with those buckets are you more excited about one of those buckets than the other are those Fair buckets to put this into so the first bucket that you mentioned uh is uh it applies across every company it's not just the crypto thing so like helping developers become more productive helping companies do better

customer engagement customer support that applies to everywhere uh not just crypto and I actually think that um helping developers become more productive that is the single most important by far the single most important use case of GPT and it's not even close I know this from personal experience last week we actually had an internal hackathon and so we spent like three days together for our offside and one of the three days we spend a whole day on an AI hackathon so all of us build stuff uh for Alliance and um it's really incredible because we only had

maybe five technical people on our team out of like 20. but um the vast majority of non-technical people uh Road code thanks to GPT and they wrote code that worked it's really impressive and there are two so more specifically there are two things um so when you write a piece of code um the first thing you need to do is architect sort of how the code works and then the second part I'm really simplifying this but the second part is writing the actual code itself um

so GPT is really really good at the second part which is you architect the whole thing you you write out the functions but you ask GPT to write the functions for you so that that in itself is a 10x Improvement and I know I experienced this personally um 10x Improvement in terms of speed of uh writing a piece of code that actually works but then there's a second piece which is called the code interpreter of uh Chaturbate I don't know if you've heard of it I think it came out a couple months ago but last week was the first time I tried it but this part it does the architecting for you so you don't even need to architect the the code in your head you can just tell GPT what you want to do in plain English

and it'll do the architect architect the code interpreter will do the architecting for you and the writing of the code for you so this this is another 10x Improvement in terms of speed so overall for me it was like 100x Improvement it's really mind-boggling it's really impressive um so uh yeah I I don't the the AI Channel let me let me show you this I uh I will share my screen here I don't know how to I've never built anything in Python I've never written a line of code in Python I wanted to build a game of pong you know the game the very simple game pawn and I look just without any

knowledge of pong I typed into gbt I said I want to build pong pull it up in vs code um I've had like one one window with gbt3 or gbt three five or whatever four maybe I forget which one it was and then another with just like the area where I was putting the code in and I just said I want to build pong I have no idea how to do it I barely all I know is HTML and CSS and a little JavaScript never written anything in Python teach me how to do it so it said here's what you do pull it up and it would give me code to put in and I would you know play around with that I'd do some Googling I'd put it in and you can see here if you can see that I think you can see my screen

right now I would say I would say look there's an error right here and it would say oh this error message is missing two required positional arguments X and Y which means this is what that means make sure to pass the correct values for X and Y for each paddle in your game and I asked it I said so should I replace this line with this line and it said no you should replace this line with this line after a couple took me about three hours maybe two hours maybe I forget one one to three hours I had a working pong game and you can see the speed here it says I just built pollen using python entirely from chat gbt whenever I got an error I would put in in chat gbt it would both explain it and fix it I know pong is basic but

being able to build this it took me all right less than one hour without ever having written python feels incredible to me and I think that's exactly what you're talking about yeah the three hours that you took would otherwise take I don't know at least a month for you to Learn Python and you know write off right right so um I think GPT is a so the more uh beginner of a developer you are uh the more uh GPT can help you um so I think GPT is less of an improvement for the experience for the 10x developers but so so GPT might be a five percent Improvement for the 10x developers but a 10x Improvement for the 1X developer and the 100x Improvement for the 0.1

developer like like like yourself like me like me no yeah I'm glad I'm a 0.1 x engineer that's great that's great crypto for the last decade has been focused on decentralizing money and like decentralizing value right like Bitcoin like decentralizing money um now it feels like there's this new conversation happening which is decentralizing AI yeah maybe could you tell me in your mind why you feel that that is important um it's important for the same reason as to why um money should be decentralized as to why a social network should be decentralized as to why Finance should be decentralized which is that you don't want these powerful models to be controlled by one one Central entity and by the way we just experienced this

where openai probably uh dumb down their models for other consumers for all their customers whereas they internally have the most powerful model still so they decide they're the king makers they decide who gets access to the powerful models and who doesn't right so you don't want to give this much power to one Central entity that's that's the whole idea behind um uh decentral Ai and a very closely related notion is privacy today when you interact with chat TPT you send them a lot of personal data and they know everything about you um

and it just doesn't feel good um so the idea of decentralized AI is to solve these two problems now there's many ways to do that to me on the inference side the most intuitive way is to what they call Edge Computing Edge AI and by Hai they mean are running an open source model on your local device so all the compute happens on our local device within an open source piece of software rather than being run on the cloud at openai and this achieves the two goals that we had before which is that one the user owns the model

because it's open source and two that the user owns their data because they don't send the data to the cloud to open AI so this Edge Computing is the most intuitive way to solve this obviously there's arguments against whether or not consumer grade Hardware is powerful enough to run the most powerful models in the world maybe we're not there yet but there are good enough of Open Source models and efficient enough of Open Source models by the way Facebook and um open source their llama model uh last last I think last week and there was a collaboration with Microsoft even I didn't read into the exact news but the the very concept of

edge computing solves the decentralization problem now there's another problem which is um which is not about privacy or self sovereignty it's more about a very pragmatic issue today which is the lack of gpus gpus are a massive shortage because everyone wants gpus everyone needs gpus to train powerful models everyone needs gpus to do inference by everyone I mean the big Tech the Casual developers that the Casual hackers to build Products off of um those those open source llm models so there's a massive shortage of gpus so so

the decentralized AI in this context tries to solve this problem by allowing the average consumer to participate in the compute Network by offering their consumer grade gpus from their laptop so for example the laptop you're working on now it's probably not being used 24 7 so when it's idle you can contribute that compute resource to the network and help the developers to Leverage The gpus to train their models or to do inference so this is a the GPU shortage problem does this make sense that feels like the

that feels like the modern day version of the 2017 file coin thesis yes yes exactly it's exactly the thesis for behind filecoin before gpus and so it's the thesis behind Falcon is for is for storage but uh now we're doing what 500 for storage before AI for for gpus yeah um I think there's a clear counter argument though that might be stronger than the benefit there which is um decentralized compute has this extreme disadvantage of being slower right in ml training because at least my basic understanding of it is

like you have this horrible communication overhead between a bunch of different types so like I've got my laptop it might be different than your laptop which is different than this person's laptop and you're there's this like communication overhead between these different Computing devices and for anyone building an AI you were saying like this kind of AI bear Market might be slow because of how quickly people are moving and I think speed is like one of the number one advantages for any AI company company building today so if you're telling me you've got maybe you can make it cheaper or whatever it is but it's gonna be 10 times slower there's no way that I'm taking that trade-off yep I'm curious what you think about it I think you respond on so so this is a very valid counter argument against uh the whole idea of decentralized compute

of or GPU decentralized GPU compute um a a very related counter argument is that the hardware that's used for training um the llm models at openai are way more powerful than the average consumer grade Hardware and there are also more standardized more specialized and so the the counter argument is basically that the consumer grade Hardware consumer grade GPU is not powerful enough to train the super large models now the counter counter argument is um maybe we don't need this decentralized AI compute to train the the openai kind of models we can use the

decentralized computer train smaller models or to or to fine-tune a a large model so it's the kind of work that doesn't require as much compute so I see the counter the the arguments and the counter arguments on both sides and overall I I think that the counter arguments against decentralized AI or decentralized compute are valid but if they're wrong they can be really wrong right as in if if decentralized compute does work it can be a really big idea because there is just a massive shortage of gpus so the payoff is very asymmetric in my

opinion I guess you could see yeah you could see a world like I think my computer right now is running this M1 chip then you know I haven't you know Apple's got the M1 and the m2s right you could see a world where like you're basically running these powerful enough chips and which are really like kind of gpus on on your on your laptop and it gets fat it gets exponentially better and better and better so yeah yeah that's interesting um so these there's decentralized compute um then there's the then there's more like the inference and like the mod the excuse me then there's like the training and like building it which is

uh listening to an interesting conversation about kind of the the chat gbts of that are getting built in China right now and how if you type in something like if I've been like what is the best political system yeah into the thing it's like oh like you know communism is the perfect version and this is the and it's like but to me that I mean that seems nuts but like if I typed in like on chat gbt like what is the perfect version of uh what is the perfect political system into this gbt4 it's gonna be like democracy because of these reasons and the reason it's pulling from that is because it's mainly pulling from you know us or english-centric sources so maybe there's like a it sounds like that's another version of decentralizing like the sources or what helped me contextualize like how you think about

that version of this decentralized the data set that is used um because it's garbage and garbage out if you feed the model training with uh very politically biased um data set then output will be a very politically biased model um so I've seen um I don't know if it was actual data or it was just opinions but the average internet is leans more liberal than conservative conservative which would make sense because uh the Liberals at least in the US are generally more educated so they use the internet more uh I think the majority of social networks are more liberal than conservatives

um and and the data the content from those social networks are what openai uses uh to train their models again this is a problem with centralization right openai can choose what kind of data they want to use uh to feed their model when I use a GPT I I clearly see I clearly feel a political bias towards uh you know liberals um very much so yeah where does crypto tie into all of this ciao because what is the are we talking about using tokens as to to enforce some of the learning here or and to speed this to exchange value what is the what is the crypto element

here I don't know if there's a crypto element I wish there was but pragmatically how to use tokens to incentivize a more politically neutral or more politically diverse content I find it very hard to do I mean conceptually yes but pragmatically feels really hard yeah yeah it feels like maybe a 2017 version of like putting you know Healthcare records on the blockchain right we're kind of trying to throw a blockchain at everything six years ago and now maybe we're trying to throw throw kind of crypto Ai and everything um so then where is the kind of crypto application here for for AI

yeah so the uh decentralized compute is something we talked about earlier but there's also um using crypto to digitally assign um content uh to fight deep fakes for example so for example this podcast that we're recording um after the recording you can sign on the ethereum blockchain uh or you can provide a signature on the ethereum blockchain that this podcast was produced really by Jason and you can put your public key on your Twitter profile that people people can use to verify that the the digital signature so this is another very

important application I don't know who should be building this I don't know if it's the job of a startup or it's the job of the big tag um but this is a this will be a very important application um especially with the elections coming up next year there will be a lot of defects a lot of you know AI generated content um to influence the um the results of the election for example so basically what you're talking about here is crypto offering like this kind of I think you tweeted about this a little bit ago that I saw offering like a checks and balances against AI right via like digital signatures to battle deep fakes and zero knowledge proofs to to kind of protect privacy here that that's the bucket that you're excited about yep so that that's the good thing

that that crypto can provide to AIS is the checks and balances but there's also a flip side which is that uh crypto is a permissionless um platform and the permissionless platform can be used by AI agents so you can think of so fundamentally AI agents are are programmable so they can leverage other interfaces to do things that they want so one of the interfaces that you can use is the blockchain and they can do this permissionlessly so so for example they can use eth as the AI native money or they can interact with uh D5 or they can play games

and if they do that they can do that permissionlessly and no one can stop them so that can get a little bit scary um So currently AI agents don't do much but if you give them the ability to do permissionless programming on chain and give them a Unstoppable form of money which is for example eth in this case I don't know what they're going to do with the type of capability so I went into the world coin offices a couple months ago um and Matt met one of their Founders not not Sam Altman but the but the other one uh Alex blania and he was kind of painting this vision of the future which said look Chachi bt4 you I mean right now you just put in a

prompt onto onto this kind of user interface and it spits out an answer he goes what gbt 6 and 7 and 8 will look like is you're giving you are giving it not just read access but right access to your computer yes so I can type in hey go go farm um go farm GMX right now just go just go just go farm GMX with 10 with 10 000 bucks and it's gonna know the perfect farming strategy it's gonna be like look there's an airdrop coming up based on all this data there's probably an airdrop coming up in in in in this month I'm gonna go farm all these tokens hear the best farming strategies and if you give it the right access it's going to be able to do that for you and if you extend that out what you could see is we think that like

there are Bots bots on Twitter right now are a problem that problem is going to get 1 million times worse it is going to grow by orders of magnitude because you're going to be able to give your AI your like little AI assistant access to your social and access like basically right access so but what it sounds like you're talking about is no what I'm talking is exactly that so when you get give AI not just read access but also write access to the blockchain things can get really interesting that's interesting I mean so the world coin this guy Alex's take on this would be that you need you need like a crypto wallet you need to scan the eyeball to digitally verify

basically like verifying who is a human online is going to be coming in in an incredibly tough Challenge and that we have to do that I think there's probably another take here though which is that there are going to be some billion dollar companies that are built on the back of embracing these Bots I don't even know if I want to call them Bots I would like just these these these AI agents and instead of trying to turn them all away they lean into them and they say look these are the new these are these are our new users basically like what you were mentioning with Nikita with Nikita's tweet it's like these are our new users like instead of pushing them away and trying to you know play whack-a-mole with Bots it's like what if we just said

these are our users now and we embrace them I don't know that's an interesting thought process that you've that you've sparked there it the embracing AI is um probably one of the two most interesting things I can think of for Consumer crypto products the other one being embracing hyper financialization um because you know trading coins speculating is is the one thing that that crypto really enables um for for the mainstream so I think um it'd be really fun for crypto consumer products to embrace that to try things that are related to speculation to hyper financialization but the other thing

that they should try is to embrace the the AI um Bots um but that I've just been thinking about this um like you know I think a lot of people are disillusioned about why you know I spent 10 years we still haven't seen a consumer product in crypto that has reached them in stream and so I tried to brainstorm ideas for just startups to try and these two things came to mind I don't know if they'll work but I think they're worth trying yeah yeah what I mean what does this look like maybe with a social network um like a web web3 social what what like maybe paint a picture of what you think

this looks like let's take it one step further yeah it's it's really hard to say yeah um I think a more a better example I can give you is um uh is D5 really uh defy summer to me is still the most magical moment in the entire history of crypto I probably sign more transactions that summer than I've ever done in the last 10 years like probably by by order of magnitude um and I think um I think a lot of people felt the same way a lot of people felt that defy summer was really a moment of magic um and I think defy summer worked because

it unapologetically embraced speculation and Hyper financialization and the whole device summer was a game it was not really Finance I mean Finance in itself is really is a game yeah it was all one big game I agree it was one big game and it was social except that the social interactions happen mostly on Twitter it was so fun right um so I kind of want to see the same consumer behavior on crypto social networks something that that feels really Degen something that feels crazy potentially outrageous so yesterday the hamsters race I'm sure you've seen it yeah that was fun

um so I want to see the stuff like weird things like that hmm I wonder if yeah I wonder if we're gonna start seeing more like social tokens kind of as well come out like you know you know what one one like version of betting basically that already happens is I find an Instagram account and I you know maybe it's I or I find an artist on Spotify and they're really smaller I find an Instagram account with 5 000 followers and I think it's really cool when that account gets to 100 000 followers I'm yeah I'm telling my friends that I found I'm like oh I found that account at 5 000. you'll you gotta imagine you'll be able to kind of bet on accounts yep and like more financialization of like tweets and I I like that tweet when it was only at 10 likes it's not viral like

I was early on that tweet I was early on that account so uh coincidentally I was actually thinking about this yesterday so there was a product that um uh came out of China in 2017. it was a product where you can as a consumer you can book time with um a celebrity and then you can speculate on the time so when you it you can almost think of it as nft except it was not nft but you can think of it as an nft that represents like 30 minutes of conversation with this celebrity and you can buy that right that the nft that gives you the right to have that 30 minute conversation and once you've bought that nft it can be traded on the open market

so if that person if the celebrity um you know commits the conversation and you know uh you know provides interesting insights to to their uh fans then you you could imagine that the price of the nft goes up over time because people will want that thing more often or they want that more uh to chat with the um the celebrity so this is another way to speculate on the value of uh celebrity or Creator Chad do you know what intro.com is yes so intro.com is exactly that without the speculation I think without the speculation yeah interesting I would love to see the

speculation because that's what crypto uniquely enables foreign talking about um AI agents and it sparked this really interesting I just like start went down this kind of Deep dive of like AI agents and one AI hiring another Ai and then you've got this kind of Master AI which is maybe like your assistant Ai and that AI hires other AIS can you explain this to me and yeah and maybe tie in like how you need crypto rails to enable that world okay so AI agents just to be upfront I don't have a strong thesis about it I don't know how it'll play out but what happened was a few

months ago there was a VC that built a open source uh you know sort of uh uh uh MVP um where you as a user you can launch a AI agent and you can tell them what to do so for example you can tell them um start a podcast right how to start a podcast for me and then the AI agent would um first research what are the what are the steps needed to start the podcast and then you know Break um the the original task into those steps and then it'll iterate from step one to step two and etc etc and then for each step it'll

break that task uh even further and then do things on our internet do more research or sometimes to have this right access that we said earlier to actually do things to transact to hire other agents etc etc so that was a really fun experiment that that the VC did but I think the hype kind of died down pretty quickly um so I don't know where it's going now it's certainly a very interesting experiment um to have these autonomous AI agents that can do things that can break tasks down for you and to execute those tasks can we Zoom way out here ciao like is this um

why do you why do you remain excited and optimistic about all of this AI or or crypto um I think all of it I mean I think there's a if someone's listening to this and they're hearing ai ai agents here's a version of someone who's listening I'm hearing about AI agents AI hiring other AI yeah social networks that are populated with millions and eventually billions of AI Bots games that aren't P2P with people but are P2P with AI it doesn't sound like a super fun world to live and I would say even for me the the eternal optimist and the the someone who just always has believed that technology enables a

better world yep even I hear this and I'm like this doesn't sound like an incredible world to live in why do you remain either excited and or optimistic about this okay so uh um where I'm optimistic is something really specific which is the 10x developer thing that we mentioned earlier because I felt it personally again I think it's by far the single most important use case of GPT it's not even close um everyone can code now and everyone can code much faster um and it's much easier for startups to launch um MVPs with GPT than it used to be that's why I'm really excited about this it

basically brings the the cost of launching products down by an order of magnitude potentially for um for developers and it's going to be a really interesting world it's going to uh I think um uh uniquely benefit the idea man right like people who have good ideas but not necessarily have the ability to code their ideas into actual products with GPT if you have good ideas you can Implement them much easier than they used to that's the most important thing I'm excited about about AI everything else I feel like it's a work in progress um I don't have a strong thesis yet

we'll see yeah yeah ciao awesome conversation man I love talking about this stuff any anything else that we're from you know either where we're at the market AI Alliance and the things that you guys are seeing in the trends anything else that we haven't that you wanted to cover that we haven't covered um I think we've covered most things um I I just think it's uh I think it's a great time to be in crypto right now you know two years ago during the bull market I felt a lot of stress it was every bull market is very stressful for me I constantly think about when to exit the market and then you as an investor you constantly think about is this person

going to be here for for a long time are they going to commit to crypto but now everything is a lot easier like everyone who's still today who's still here today talking about crypto you know they're committed to crypto yeah um it's um also uh more it's for the most part A contrarian place to be and I feel very comfortable being in an environment like this yeah yeah I agree with you I was at the office with a bunch of our team uh week ago two weeks ago whenever the Ripple SEC thing happened and um my first thought was pure Elation and excitement because that's a that's one of those moments that can bring you out of a bear Market I think is that Ripple SEC thing

my second thought was frustration that the bear Market was ending just wanting to hang on a little bit longer for this Blissful time in the market where Everything feels so simple and knowing that the chaos will before we know it be here so I I share your thoughts yeah awesome man well wishing you guys all the best I will uh and I will talk to you soon I'm sure we'll have you on again soon thanks for the time though thanks [Music]