Web3's Next Level: DePIN, AI, and the Future of Automation

Transcript

Good morning everyone. Excited for you all to be here. Appreciate you coming out this morning. Last day of the conference. Exciting weekend con. Very very hot. Um and uh I'm very excited today to talk to you about where we see the industry headed. Uh and share a journey with you of uh where we've been uh throughout the last eight years in this space. Um, I can't start at the beginning because we don't have enough time, but I'll start nearly at the beginning. Uh, when my co-founder and my founding team and I were up in the mountains in California and Tahoe uh doing our I guess technically one of our first team retreats. Um, and uh at a time where DeFi was was starting to come alive. This was after we uh lived

through the ICO era. uh lived through the era where we call uh 2018 and 19 where uh we were in a dark room, lights were off, there was no exit, we didn't know where we were going. Uh neither did the industry. But luckily, DeFi came alive and uh we started to plug into some really interesting technology. Uh this is when uh unis swap started to come online and we started to see some new interesting contracts that didn't just swap tokens for other tokens but actually had kind of some yield strategies, liquidity pools, factory contracts. And one of the bigger questions we asked once we see saw these things come online was was how do you kind of focus on speed as a trader? How do you get ahead slightly as an edge?

We've obviously seen a lot of that uh prior to the crypto space and the high frequency trading space and other traditional markets. But uh when you're looking at onchain activity, could you create more intelligent systems? And so one of the first things we did in DeFi uh was to build out this little bot on Telegram, very simple, humble bot called Unispotter. Uh one of the first bots for unis swap at the time. And uh we just published a little blog post saying, "Hey, we're listening to real-time activity on chain. uh we can move uh pair pool liquidity and and layer in pricing data for day traders to get a little bit of alpha, get a little bit of edge on the market. Uh we published this in a simple blog post uh and we had over 7,000 people sign up for this bot in the

first 10 days. Whoa. Okay, people want this. People don't want to work for their data. uh and traditionally before this you know you're cross referencing onchain data with block explorers looking at some price data uh through some pretty simple uh APIs or other data sources at the time and you could not do that under a second we could easily do that in real time. So uh we saw the value and and at this phase we really focused on speed. Uh what was fun after this uh era of the early beginning uh start of um unispot uh unis swap uh was we entered the the really fun phase we all know and love that is defi summer.

Uh if any of you have seen severance great show I recommend the guy dancing like that that's DeFi summer that's the vibe that everyone felt and when we look back we feel today. Um and when we entered this era it it just wasn't unis u unis swap anymore. uh we had other forks of unis swap, we had new exchanges coming online, we had lending protocols and uh yield bearing strategies that were all starting to take shape. And when we thought about that era, it wasn't just about speed. Uh but it was more about automation. And uh we evolved as as well internally and we wanted to help more people automate and build bots like unis uh unispotter uh themselves and and provide that level of tooling.

And so we built out a new automation system uh called dispatch. And uh it was a more smarter uh intelligent system that could react in real time to onchain activity. And um this was a really interesting product. And and what was interesting throughout this time DeFi summer came uh COVID came uh and then NFTs came. Uh who knew that sequence was going to play out but it did. And uh what was interesting the way we structured this product and the way we thought about the industry we knew DeFi was one of many verticals to come in crypto. But we didn't know what the next one was, but we wanted to build a system that was intelligent enough and reactive enough that you could drop in any smart contract. Listen to smart contract events and functions regardless of the standard. Uh, you know, obviously we're looking at 20 tokens, factory contracts.

Now we're looking at ERC 721s and 1155s, but our system was agnostic in that regard. And so we started helping out teams beyond DeFi. Now, uh, we had teams looking at uh, they were launching games and doing in-game minting of of tokens. Uh we saw some early really really interesting use cases I I like to say like early RWA use cases uh not of massive assets but of uh authenticity for brands. Um we were working with a group that was working with F1 teams and um uh Insomniac Games on Spider-Man drops for specific authentic apparel uh which was all very interesting. And um as this era came and all the excitement came and a lot of dollars flow flew um rushed into the space uh I always like

to think that era was longer than it was but it was quickly cut off. Um and I'm not going to spend too much time and kill the mood of the room. Uh but you know we we all felt we all knew what happened here. Um and in reality for us uh we kept building. We've lived through I think the 18 and 19 era was probably the worst time. This wasn't great. uh I had some bunch of experience I'm happy to share you know outside of this room but uh as we continued down this path um and the market came down uh a new things were happening behind the scenes and what we like to call the great unbundling um so obviously there's a crash there's a moment uh and then we moved into an era where it wasn't just about the top tier chains uh but it was

about rollups it was about L1's L2s L3s L4s totally new technology um 1 million transactions per second. All exciting stuff. Uh but as in from an infrastructure standpoint, we had to kind of rethink what how we handled it. And so for us, it it felt like that Oprah moment uh for those that know Oprah uh where everyone on reach under your seat, you get a blockchain. And um and uh it was exciting. Uh it was interesting, but we started instead of us getting protocols reaching out that were on Ethereum or Polygon or Avalanche, we had net new blockchains reaching out to us saying, "Hey, can you index our infrastructure? Can you hook up dispatch to it? And while we would love to do that, the realities of, you know, indexing OP stacks that are doing

a terabyte of data per month is that that's expensive. That's not just traditional archival storage. This these people are starting to require NVMe GP3 style storage in AWS, which is really expensive. Uh, as you start looking at the node providers, they have to offset that cost as well. And and you know, you're getting you have this meter that's continuing to rack up. So, as we dug into this area, our infrastructure cost went up to the right. And that's not the up and to the right you want uh typically. And so, we had to figure something out and uh we had to start thinking more holistically. And what we see from the great unbundling to the great funneling and uh quick side story uh before we got into crypto, my co-founder and I have been working together for 13 years. And uh we started

and scaled another company in the higher education space. As we sold that company and transitioned into crypto, uh this was at the time Microsoft Azure was getting pretty competitive with moving into the cloud space. They gave us just under half a million dollars of credits. We were really excited. We were just starting off in this new project. We started running some Ethereum nodes and we were immediately sent to cease and desist. Okay, what's going on? Um they said, "Hey, you can't you can't mine cryptocurrencies uh using our infrastructure." We were running an RPC node. They had absolutely no idea what we were doing. uh and that is what it is. But what was interesting at this time as we kind of entered this next phase of growth within crypto and all these new chains came online. I started seeing a shift as we saw our costs go up

on our traditional bills. So did the uh ops people and strategy people at GCP AWS and Azure and they started showing up in crypto and they were on stage with some of the top protocols announcing partnerships. so exciting and to see these big companies uh partnering with these small companies. And don't get me wrong, these they fill a very important void in getting people off the ground. But when I started in 178, everyone was talking about decentralization. And here we are literally every single protocol in some capacity, leaning on the most centralized infrastructure in the world. Um and as we started unpacking that and realizing that our bill was going way up, everyone else's bill was also going way up. And the reality of credits, while they're nice, in the beginning, they run out. And when we spoke to most

founders, every single person's biggest line item on their expense sheet beyond salaries was cloud spend. And so I was like, how do you actually tackle that problem? So we took a turn. Uh we went back to kind of some old school stuff. Uh back in the day, um little story here, my co-founder before I started working with him a long time ago, uh his head of DevOps and him, they got acquired uh by this big pharmaceutical company where they were running a massive on-prem infrastructure. uh postacquisition after it was all said and done uh the they got to retain all of that hardware and so this was sitting in his garage for a number of years and when I finally got around to hiring Kevin the one standing up on your right um he said hey I'm managing AWS I see

this bill it's growing we're going to do everything we can to cut costs but in reality we're going to onboard all these chains it's going to continue to go up I have this hardware in the garage collecting dust are you interested in spinning it up I was like well what would it take he's like well I don't enough power in the house to actually get this online. Uh, but I could, you know, slim it down to one power supply, uh, get a 50ft extension cord, unplug my dryer, and plug in the rack. Sounded reasonable. So, I ordered him a 50ft extension cord. We plugged it in, uh, and we had a home lab going for a few months, and the test there was to, do we have enough compute, even with this legacy hardware, to offset our internal cost of at the time was nearly a quart

million dollars to spend per year. And uh as we did that uh we quickly realized we had more than enough capacity and that even though that this was legacy hardware the responsiveness of it as we started kind of re uh getting into these boxes and starting standing up our own infrastructure was substantially faster than that of AWS. Um so we started digging in a bit more did that in the home lab like okay let's upgrade some of the hardware world to move it to the enterprise data facility. Ironically the same facility they got acquired in years prior. Um and we racked up the equipment about several months back. uh and hooked up our own direct ISP drops uh you know power supply all the fun stuff that comes with you know stacking and racking hardware um and as we looked around we said you know who else is here like this

this data center is at full capacity uh you know we're under the impression everyone's in the cloud uh we see some of the biggest sports networks we see some of the biggest streaming companies we see a lot of infrastructure where it's like the realities of scale is that you're not typically sticking with the hyperscalers at some point or you have a hybrid strategy and so as we dug deeper into the data center space. One of the most interesting things we found uh after you know going down multiple rabbit holes of modular nuclear and all this stuff is that when we're talking to boutique cloud providers, when we're talking to data centers uh and traditional bare metal groups, they all sit on idle capacity. This is capacity that they have as additional headroom in case there's a spike. But in reality, a lot of this just sits waiting. And as we

approached these groups and started talking to them saying, "Hey, could we acquire some of the hardware or would you be open to, you know, lending some of this hardware out as as we become more of a channel for you to increase utilization rates?" They were all on board. And so we had to figure out a way well if there's that much idle capacity if there's a way for these groups that want to set their own pricing on their own compute uh but we can go from you know a thousand cores to 100,000 cores across our CPU disk and RAM stack um how do we make that in a way that's approachable accessible and affordable uh for a lot of these groups and for us our goal wasn't to cut cost by 10 20% that's not going to move the needle for folks people aren't going to move their

production workload off of AWS for that we're moving at least 50% cost reduction across the entire enterprise stack in production. And so um we started building up a a pipeline of different groups that wanted to add capacity. And um that was really exciting. And and as we got to that point, we had to realize, you know, how are we going to coordinate it all and what is that next step? And that's where we get into kind of web 3's next level, uh we like to call it. And this is a culmination of all of the things that we've been working on for the last eight years. Um so it started in DeFi uh really focused on speed um and that kind of piece of you know meeting the market where it is and I

still think DeFi is one of the best product markets I think most of us can agree within crypto. I think the compute has been this rapidly growing space that we've stumbled upon out of necessity uh to offset our own infrastructure but to also scale to a different level um and to maintain reasonable costs and pass those cost savings along to other users. Uh and then the biggest piece that we think is going to be the biggest unlock not just in web 3 but across the globe uh are a agents in a fully autonomous agent-to- agent economy. And the only true way to do this in our opinion is to have a purpose-built kind of core. Um and so underpinning all of this is a new uh layer 1 EVM compatible L1 blockchain that we've been building. Um that is

going to be the coordination layer, the settlement layer um that all of these systems can run upon. uh the rails that it can run upon and it's not competing with memes. It's not competing with other use cases. It's purpose-built in a way for these systems to run. Um and we're super excited about it. It's called Flora. Um this is the first time I'm publicly talking about it today. Uh soft launch of sorts. Um but we're excited to share this with the world. We're excited to start building uh this in public uh with you all. And this is something that you know this is eight years of us in this space. uh we've been wanting to do something big in the more public domain for a while. Obviously, we've been building SAS products and infrastructure. Um but we've wanted to

build something that was more open and a kind of a public good and uh floor is going to be a a big uh project for us internally. Um and something that we're excited for for many people to follow along as we continue this journey. Um so yeah, we're ready for you know, call it AI summer or D FI AI summer. I don't know how you even pronounced the defi in different terms, but this is us today. Um, you know, we're here. This is us tonight maybe um or yesterday or the day before, but uh within the years to come that beach will be beyond full and we're going to outgrow Khan. So, how are we going to do that? Uh we believe Florida is going to be a big part of that and uh we're excited for you all to follow the journey. If you want to follow along,

feel free. You'll be like literally the 15th or 20th per person on our Twitter if you want to follow us over there. Um so, it's early days and and we're excited to to announce it here at the community conference. One of the big things for us is trust uh and and kind of building in the open and building out a community. Um something I've always loved to do across my last startup and and and this one too. So we're excited to uh to unveil it today. So thank you all for your time. Happy to answer any questions. We got a few minutes. Wonderful.

Go for it. Why do you think uh oh thanks um what do you think has been I guess the biggest reason that AI isn't yet a huge part of web 3? Yeah um good question. So I think one thing with AI right now um obviously it's AI's uh been around for a while in different form factors but in terms of the big unlock in the last 18 or so months has been like this new uh chat interface or prompting interface. Um, I think as we think about web 3, the use cases aren't just gonna like prompting is not the endgame for crypto uh or for AI uh and the use cases for crypto. I think it's going to be more automated agents that are running behind the scenes doing different type of strategies across DeFi. Uh we also have some interesting like DevOps style uh

multi-agent workflows that are going to be running against the compute stack um that are already helping us increase like four times capacity. I think the issue when you think about that level of integration across a multi-step agent workflow is that some people are doing that today but not the majority. And I think that that interface needs to change. Um so it's not just about custom prompts and get, you know, slight efficiencies. It's about full-blown automation across that stack. And we believe the marketplace we're starting to build in Flora um will be templatized orchestration tools like that that you can click and deploy. So like no code, low code, low op style pipelines that is accessible to more people um for more powerful style agents.

Yeah. Um two questions. Where are you guys based? And also um specifically with like the use case for around DeFi, how do you guys um intend to launch? Sure. We're based uh all over the US right now uh in terms of the core team. Um, and then in terms of DeFi, so one of the, so we're talking to a lot of existing DeFi protocols. We're not trying to necessarily reinvent that wheel. We just want native DeFi rails there. Uh, talking to some of the biggest stable coin groups as well. So we want the ability for agents to transact with one another seamlessly. We don't want to force bridging across all these agents. I think as you start thinking about real use cases and within the crypto space, the set of privacy, trust, and concern there can become potentially an issue. Uh so having it

all living under the hood here with again the other reason we want EVM compatible all the top D5 protocols are already there and so being able to onboard them is going to be quite seamless. Um the other piece that we're uh considering and we're still exploring is uh opening up the marketplace to other top tier uh blockchains. So uh we're not under the impression that we force people to use Flora. Flora is that settlement layer ultimately underneath the hood. But um we want to have that marketplace live across other ecosystems so people can use other D5 protocols connect agents to other workflows and all that stuff but behind the scenes Flora will settle back to the the core chain. So is the idea that the agents live on is the idea that the agents live on the Flora network and then um and then would you guys have the users or

would the users come in through other places? Yeah, users can come through different places in terms of the different chains but ultimately the agents would live within smart contracts initially on Flora. um that can also be compatible across other chains. Got it. Thanks. Yeah. Awesome. All right, everyone. Thank you so much for your time. Appreciate it. [Applause]