**A** (0:05):
Hello, everybody. I have an amazing founder here again today, Sunny from Osmosis. Sunny, thanks for taking a moment to check in with us today.
**B** (0:13):
Thank you for having me.
**A** (0:15):
So I want to hear everything. But we have in this crazy event, Ethan, we're maybe 15, 20 minutes together. So I want to really quickly get into your origin story. How did you take the pill into crypto?
**B** (0:29):
Yeah, you know, there's many stages of, like, getting into crypto. It's like, when did you first hear about it? When did you first buy your first bitcoin? When did you go, like, jump in full time?
**A** (0:40):
Yeah.
**B** (0:40):
But I guess, you know, first time I ever heard about crypto was, I think, senior year of high school. A friend walks up to me and says, dogecoin just sponsored the Jamaican bobsled team. And I'm like, what does that mean? None of that makes any sentence made any sense.
**A** (0:57):
Yeah.
**B** (0:57):
And so I had to go home that evening and look up what is dogecoin? And I'm like, oh, that's interesting. And then obviously, it talks about bitcoin in there. So I'm like, okay, well, let me look up what is bitcoin? And I'm like, oh, that's really interesting. And so that kind of was in the back of my mind for a little bit. And then, you know, I get to next year.
**A** (1:17):
At that point, I presume you didn't know what bobsledding was.
**B** (1:19):
I did know what bobsledding was. Cool Runnings, I think, taught everyone what bopsledding was.
**A** (1:23):
And Elon was not a fanboy of.
**B** (1:25):
Dogecoin at that point. No, no, no. Dogecoin was just like, what weird online community of people who just like to tip each other on the Reddit. Yeah. So I just had that in my mind. I get to UC Berkeley the following a couple months later as a freshman, and there's, like, a lot of clubs and stuff. You're finding out what you want to do. There's this, like, bitcoin club. And I'm like, oh, I remember looking up bitcoin. I was, like, studying computer science and political economy. And so it was a very obvious interest of the two. So I joined that club was very over my head, but I'm like, oh, wow, I'm. These people seem very smart, so I'm gonna keep showing up. What we started to do was me and two friends, we decided to teach a class on bitcoin and, like, how the technicals work and everything, because for me, the best way to learn something is to sign up and teach it.
**A** (2:14):
And you are equipped with the opportunity to do that because in Berkeley you can as a student teach classes, correct?
**B** (2:20):
Yes, exactly. Yeah, we have this cool program called Decals. As long as you get like, you need a professor to support it and back it. So we had Professor Dawn Song, she works on Oasis now, but yeah, she was sort of our backing professor. And yeah, we had a lot of students join. I kind of like was learning while teaching, but it's, you know, you can get away with that. And yeah, with those students, we created a new club called Blockchain at Berkeley. Kind of grew that did. Did a lot of like educational and research stuff, a little bit of consulting.
**A** (2:51):
Work and kept offering classes along the way. Kept offering classes. A class on Rust was first offered in Berkeley. Right. As a result of your work there.
**B** (2:59):
Yeah. So it's like really cool. I think that education was like one of the most important things of like what Blockchain at Berkeley did. Okay, then let's fast forward to like summer of 2017.
**A** (3:10):
Okay.
**B** (3:10):
I just started getting interested in Ethereum a little bit. This someone convinced me to check it out. So I started interning at Consensus just as a way of learning about. More about Ethereum.
**A** (3:20):
Which domain of Consensus were you at?
**B** (3:22):
It was basically like a proto version of the graph. Yeah. So we were like building like a subgraph system for Ethereum basically. So I was doing that. But I don't know. Something I just never. Something about Ethereum just never clicked for me. I was still a little bit of a bitcoin maxi, I think, and I was like, oh, this Ethereum stuff seems really cool, but I want to bring value to Bitcoin and do cool stuff on Bitcoin. But while I was at Consensus, I got super interested in Proof of Stake. And me and two friends, another friend, we just started reading every white paper we could and started listening a lot to the Epicenter podcast. And I was like, oh, wow, learning so much from this.
**A** (4:00):
And what was the impetus for the proof of stake? Was it the idea that you could own something and it would be a passive income generating asset or what was the appeal?
**B** (4:10):
I think it was that you have more levers in which to design like mechanism designed incentive systems. Because in proof of work you have a cost. So there's a cost to doing things right, but there's no way of punishing bad behavior or I guess, no, say the other way around. There's the same cost for both good and bad behavior. And the idea is that the cost will be high enough that like, oh, we can, you Know, it'll only incentivize the good behavior doing the bad behavior. It doesn't get accepted.
**A** (4:43):
Yeah.
**B** (4:43):
Nice thing with proof of stake, you have this, like, whole slashing concept. So you can make the cost for good behavior pretty low. There's like some capital cost, but you can make the cost for bad behavior very high by slashing. And it's like, oh, wow. With this, we can actually. That's a lever we can do more interesting mechanism designs with. Right. So that was the original impetus for me. And the idea was like, okay, proof of stake allows you to do this with the blockchain, but then you can have the blockchain validator start to do more, maybe provide oracles, provide some privacy functionality, npc, all this sort of stuff. And the slashing acts as the stick.
**A** (5:21):
Carrot and the stick.
**B** (5:22):
Yeah, but proof of work only gave you the carrot. Proof of stake gives you a carrot and a stick.
**A** (5:26):
Yeah. Also made it, like, hard to get carrot.
**B** (5:29):
Yes.
**A** (5:30):
Implicit every time you tried it or some punishment. But yeah, more direct. It's like incentives and taxes.
**B** (5:37):
Exactly right. We were reading all the proof of stake papers, and I really liked Tendermint because I thought, like, wow, this is very simple and elegant almost, and we can do it quickly. And I like the vast finality function of it. So I reached out to the Tendermint team and they told me about this Cosmos thing. And I didn't realize when I reached out that, oh, that was like the same team working on Tendermint and Cosmos, but. But they started pitching it to me and I'm like, oh, this actually fixes a lot of my issues with Ethereum because, like, I really, like. They pit this whole idea of, like, a world of many app chains and, like, bridges between them. Yeah, a standard. Yeah. And at the time, Ethereum was very much like, you know, focused on like, oh, we're gonna have this one chain to rule them all. Everything's gonna be built here. And I'm like, for me, it was like, oh, well, bitcoin is an app chain. It's just for payments. And if we can bridge Bitcoin off, then we can build applications just like the Ethereum people are. But we can build applications around Bitcoin. Bitcoin will be the asset, that will be the money of this whole ecosystem. And that's, I think, what really excited me and why I kind of quit my internship. I had consensus halfway through the summer and instead started working with the Cosmos team. Really enjoyed what I was doing. And then basically, come fall, a day before my Classes started, I dropped out to work on Cosmos full time.
**A** (6:58):
Yeah, I see. So summer of 2017, what's the gestalt like? Tendermint Cosmos. You're there. Coming from an ex intern from Consensys. What's the ongoing discourses? What are you guys trying to build? What are you guys thinking about that time?
**B** (7:15):
Yeah, so at the time I would say like I originally came in with like a lot of ideas for how to improve the Tendermint consensus protocol. You know, that code base was mostly done. I think the biggest thing that when I jumped in, that's when we just started building the Cosmos SDK. Because the Tendermint consensus engine had existed for a while that was originally created in 2014, but now it was like. And it allowed people to build any state machine in any language they want on top of it. But what we were realizing was like, well, building entire state machines from scratch is hard. People don't want to write their own dk. Yeah. People don't want to write their own cryptography libraries and hashing function on databases. And so let's go in SDK give a bunch of good same defaults, standard implementations of the things that everyone will want, staking, tokens, governance, all this sort of stuff. But you can make it, you can modify it, right. You can go. You don't have to use the defaults that are there.
**A** (8:12):
Yeah.
**B** (8:13):
And so, yeah, so when I jumped in, I mean, to be honest, I was probably not the most helpful when I joined, you know, I was like barely, you know, just a second year college student. I think the Tenderman team, one thing that they were really good at is just like finding young talent, I think. And so, you know, they just. It probably took me six months, six to 12 months to actually get to the point where I'm actually being contributing back and being helpful. But you know, I like that they invested that time in me and other people as well, including my co founder of Osmosis was also part of. He was an intern at Tendermint the following year.
**A** (8:43):
Yeah.
**B** (8:44):
But when we jumped in, really just focusing on the Cosmos SDK, that was the number one focus. And we were trying to ship the Cosmos Hub blockchain.
**A** (8:51):
I see. So your first core product was the Proto Graph in Consensus then working on the Cosmos SDK. Yeah, so that's super helpful. Then let's fast forward a little bit to Osmosis. You get going with Osmosis. What's the vision at the start?
**B** (9:09):
Yeah, so Osmosis started where the first thing that we started with Was we really wanted to do something with privacy. And my co founder Dave, he's one of privacy experts, Right. He did a lot of research on zero knowledge proofs and stuff. And so we're like, okay, well with privacy, what is the most practical thing? That's like the biggest issue right now. And we saw like front running. Front running on Ethereum was getting really bad. It was just not as bad as it is today. But it was just starting to kick off. I was very close friends with the Flashbots team and so we all were sort of researching this whole MEV issue together and I wanted to figure out, okay, how do we solve this and realize that, okay, well the fundamental problem here is there's a privacy breach happening that everyone can read everything in the mempool. And so if we can bring privacy to the mempool, then that's how we can solve this. You can't front run if you can't read anywhere.
**A** (10:02):
Exactly. You can't make a sandwich if you don't know what spread and what's.
**B** (10:05):
Exactly.
**A** (10:06):
Yeah.
**B** (10:06):
And so we looked at all the different solutions for how to build privacy of the mempool. There's three main ones that we realize are possible. There's SGX based solutions, which has its own set of issues. Time lock encryption, which has its own set of issues, and threshold encryption.
**A** (10:23):
Yeah.
**B** (10:23):
And out of the three we like, we're like, okay, threshold encryption seems to be the one that makes has the least drawbacks and feasible right now. Yeah. And so we started building and architecting and building a threshold decryption system using a tendermint chain. And then we're like, well, what's the application? This is a feature, not a product.
**A** (10:46):
We need something in which we will not allow front running.
**B** (10:50):
Yes, exactly. What should we build? Well, actually, funny, before we did, we actually originally were like, at the time it was hard to explain because there's no terminology around it. But now the terminology would be a shared sequencer we were trying to build. We were like, oh, what if we build a chain that does threshold decryption and then it allows rollups to execute based off of what's being threshold encrypted Here we were a little bit skeptical whether people would use it or not, whether rollups wanted to use that versus just doing it internally.
**A** (11:20):
Internally being in the layer one in the rollup. In the rollup.
**B** (11:24):
Yeah. I guess we expected that the rollups would be more decentralized than they are right now. Like not these single sequencer ones. Rollups would still be running A consensus protocol like Tendermint and then still using the rollup system. And we thought that, oh, the rollups are just going to do threshold encryption within their own thing. They're not going to rely on some shared sequencer sort of system. But we were like, okay, let's at least we'll validate it while we're building. Meanwhile, at the same time this is all happening on the other side of the world. Josh and Tony, they were the creators of the Kepler wallet, which is the main wallet for the Cosmos ecosystem. Yeah, they were not sure how to monetize a wallet. This is pre metamask swaps and people realized, oh, wallets are actually extremely profitable. So Tony started at a hackathon, built the first version of Osmosis, which was a dex on the Cosmos SDK. And we were connected to Josh and Tony, Dave and I about like, hey, why don't we help them out and help them launch this? And it'll be our first, they'll be our first customer for our sequencer system. And so we just started working together and then we're like, after working together we're like, wait, why don't we just completely just merge? Why are we building the shared sequencer thing at all? Why don't we just like focus on the decks completely and just build a more user facing product rather than this?
**A** (12:44):
Like so how do you go from the sequencer and wallet and combine them to in your cognitive map get to a dex? Like what is the synapses that are crossing over that. I'm not registering that.
**B** (12:56):
Yes. So basically what happened was so there was a company in Korea, Chain Abscess, that built the wallet and then there was Sika that we were built, which is our validator company, but we were building threshold encryption and then we started Osmosis as a joint project.
**A** (13:09):
Yeah.
**B** (13:09):
So two companies each have two products, but one of them was a joint product. And so yeah, that's sort of how it started. We eventually, it's like cliff our threshold encryption stuff, merge it in chain. Appsys still maintains Kepler as well, along with being a core contributor to Osmosis for us it was like, this makes sense because like I said, we were worried about would rollups want to use the shared sequencer. So like, oh, why don't we build our own product so that we. I think being as close to the user is very important.
**A** (13:40):
Yeah, I hear you. So you went straight to the dex. What was the vision for a successful dex at that time and how that has sort of evolved?
**B** (13:49):
Yeah. So like I said when we started it was like, oh, we want to build the privacy focused decks.
**A** (13:54):
Yes. No front running here.
**B** (13:55):
No front running. And then also other sorts of privacy as well, like on chain privacy. Because today every Dex, everyone's leaking their entire trading strategies. Right. And that's like, not okay. Yeah, so that was the original goal, privacy focused Dex. But at the same time, like Cosmos IBC was just starting to come active and we're like, wait, there's an opportunity here. Because what we noticed was Cosmos ecosystem assets were like very. There's a lot of Cosmos ecosystem assets.
**A** (14:27):
Sure.
**B** (14:28):
And they were very high value. Like, I think even today probably, I think the Cosmos ecosystem has the most number of assets in the second most number of assets in like top 200 by market cap, other than Ethereum. But at the time, there was very hard to get these assets listed on centralized exchanges because unlike ETH assets, ERC 20s, they just, you know, centralized exchanges have to integrate one blockchain, just add ERC 20s.
**A** (14:52):
Yeah.
**B** (14:52):
But for every Cosmos asset, they'd have to integrate a new chain. And so it was harder for Cosmos assets to get listed on centralized exchanges. And so we saw like, wait, there's a lot of demand for these assets, but not a lot of like, you know, there's an untapped market here, basically.
**A** (15:06):
Yeah.
**B** (15:07):
And so we're like, well, maybe we can launch Osmosis and tap into this demand and be the first killer Dex as soon as IBC goes live. And so that's sort of what happened where we were right. Yeah, we were right. And so we launched and got a lot of traction that way. And then once you've launched a product and everything changes because, like, you are now dealing with the demands of the market. I mean, Osmosis was very successful in its early days, I guess you could say we got product market fit from day one of launch.
**A** (15:42):
Then all of us, like four years of hard work and overnight success.
**B** (15:44):
Yes, exactly. But yeah, so then from there it was like, you know, kind of figuring out the needs of the market. You know, I guess one thing is like, we got very sort of sucked into the Terra supernova effect. Osmosis was like the biggest decks for ust. And so, you know, that was a market that we're like, oh, wow, this is a fast growing market we're very well positioned to tap into. That came back to bite us. But, you know, I think the whole Terra, you know, we can talk about that if you want, but I think it was actually, in the end still very positive for Cosmos As a whole.
**A** (16:17):
It showed like, hey, there's a lot of resiliency.
**B** (16:19):
Yeah, yeah. And it's like all the developers and everything, they still stayed in the Cosmos ecosystem. A lot of them are building on top of Osmosis today or just building other Cosmos chains still. Our end goal is still to push for a lot of this privacy stuff. It's still like in development. As we started developing it, we realized, oh, there's actually a little bit more stuff that we need to do, like, oh, we need to add this functionality to the Tendermint consensus. And like. So. But yeah, was it a challenge for.
**A** (16:44):
You to think about? Okay, these are the features we're going to do super Fluid staking. We're going to do account abstraction sessions, what have you, and then trying to prioritize or dedicate specific talent to them. How did you realize what you need to do and then actually get around to doing this in a reasonable timeframe?
**B** (17:04):
I don't know if this exactly answers the question, but one of the challenges that we often face is how do we prioritize things that we find intellectually cool and innovative versus things that are direct demands from the market. Super Fluid staking was, we were really excited by it because we're like, this is an innovative thing that no one has done before.
**A** (17:26):
Yes.
**B** (17:26):
And is it the most like, important product facing thing? Maybe it's nice because it gives LPs more yield from staking, so you don't have to choose. But at the end of the day, it's not like this makes the trader experience like 10 times better.
**A** (17:40):
Right.
**B** (17:40):
But I think what was important for us was we wanted to, you know, this is like back when like all the projects you see are just like forks of the same three code bases. Right. And we wanted to really set ourselves apart and be like, no, look, we, we are actually able to do innovative.
**A** (17:54):
Things and that really built the narrative for how. Because you're on app specific chain.
**B** (18:00):
Yeah. There's an example of we were able to show this is something we can only do as an app chain and no one else could do this. And then there's other things that we've been working on. Another one example of that would be one of my biggest qualms with every existing blockchain today is you have to have one token to pay your gas fees in. And this is like just the worst UX and it doesn't make any sense for Osmosis because as a dex, people are bringing in other assets. Let's say you bring in Atom or ETH or usdc. You don't have any OSMO yet. How can you pay? Yeah, it doesn't make any sense. So we modified our chain to allow transaction fees to be paid in any token.
**A** (18:38):
And guess what, there's a Dex running there. You might, you know, if you really, really needed to get it to osmo.
**B** (18:44):
Exactly. So we're able to use the OSMOSIS TWOP price as an oracle for how to weigh these assets. And then we collect all the fees on some regular cadence. The protocol just sells them all for osmo. So it still provides OSMO buy pressure effectively and then distributes it to validators.
**A** (18:59):
Yeah, I hear you. This is pointing at the sort of tension between intellectually curious, maybe good marketing versus high value features and how to prioritize them.
**B** (19:10):
Yeah.
**A** (19:10):
So if you're faced with the same situation again in the future, how would you approach that? Like, what's your takeaway?
**B** (19:16):
One takeaway in general is like, know when to like drop a project. Like there's oftentimes things that we would. Part of this is tension that we always have, which was like, we. There's a lot of projects who like say they will do stuff and then like don't ship it or like the timelines never make sense. Like for superfluid, like we were like, we want to say a date, we're gonna launch it and actually stick to it. And back then I basically said publicly I would shave my head if we don't ship it by that date. And it was really nice that we, I don't know, it was important to us to be the team that actually ships when we said we would. But sometimes that also turns out to maybe not be the best choice where. So, for example, Osmosis was building stable swaps and we started building the stable swaps back in April, back when UST still existed. And we're like, oh, we will. You know, we were already the base decks for ust. That's like, okay, let's make sure we have a stable swap so we can be like the UST USDC primary like pool in crypto. And we started working on that due to a number of things, including Terra collapse and other stuff we had to deal with. The development of that took a lot longer than we wanted and we shipped it in November. But like, it didn't have much product market fit at this point because in Cosmos there's currently one popular stablecoin which is only usdc. And so it's like there's no real product market fit for this stablecoin.
**A** (20:43):
To swap it for anything else.
**B** (20:45):
Yeah, exactly. And also it's like concentrated liquidity is also. We are concentrated liquidity implementation is also just like right around the corner. And it's like, to be honest, a little bit unclear what the benefit, like how important a stable swap protocol is when you have concentrated liquidity. Because if you look today, actually uni v3 is actually quite competitive with carv even on stable to stable transactions. So I guess one thing that one lesson that we learned was at some point we were very much in this mode of oh, we said we're building stable swap and we are going to make sure we ship that because that's what we told the community that we would do. But like, maybe at some point along of that we should have like looked back and reevaluated, especially after the terror collapse. Like, wait a second, how important is stable swap? The market? Clearly something about the market conditions have changed in a meaningful way. We should have reevaluated that point.
**A** (21:33):
Okay, but now it seems like you're in a better position as a leadership group to say, well, we always do what we say. Look, we did it even when we realized it wasn't necessarily the best because. Because conditions have shifted now. It gives you cred to say By April of 2024 we committed to do this. But January comes around and is like, you know what? Crypto prices are gonna skyrocket. This is not the most important thing right now. Let's do something else. And then I think it was still a good call to.
**B** (22:05):
I think there's trade offs.
**A** (22:06):
Right?
**B** (22:07):
Yeah.
**A** (22:09):
Is there some pain point that took you three to nine months to actually address? Find a solution to that you wish you knew before that you'd like to share with other founders?
**B** (22:19):
Yeah, I mean, so I think, I think that prioritization of like, you know, make sure you know when to like put things on pause is very important. Another would be like, definitely make sure you're hiring like the best people. So I think maybe like we, looking back on our experience from Tendermint, you know, like I said, we brought in a lot of early like very young people, but I think we had enough like more senior people as well that we were able to support that. I think with osmosis for a while we had too many junior people, not enough very senior people. And that became like an issue. So I think it's like, make sure that you're hiring senior people along with like making bigger bets on these more junior people that you might have a lot of conviction in. But like Without a support system and mentoring system and a way for them to learn, they're not going to be set up for success. So make sure you're balancing those.
**A** (23:15):
Right. So speaking of mentoring system, what sort of good templates? What's a good way to do it?
**B** (23:21):
So to be honest, we're in this process of learning this right now. Right. I think, but like making sure that people have one. You know, there's a senior dev that is able to like answer questions for more interns or junior devs and stuff like that.
**A** (23:36):
Yeah.
**B** (23:37):
So and I guess another thing we learned was like separate learning is like we're a very remote company. Like, you know, like I said, we started off Korea in the U.S. right. Yeah. @ this point I don't think we have more than three people in any one city. So we're just like completely geo distributed. And this.
**A** (23:52):
No meteor can take out osmosis, no.
**B** (23:54):
Media can take us out. But like, and it's nice you can hire amazing talent who live anywhere. We have people in Italy, in Sweden and Korea, wherever. But it definitely makes it harder to build like proper or company culture and stuff. And like, and so I think it's important to do regular team meetups even if it's for like smaller like team meetups. Like for example, we just got back from Taiwan where we did a front end team meeting and just like, you know, getting that like synchronous, like just knocking things out. And also another one was we did like the chain development team met in Berkeley for an entire month and just like, you know, not everyone has to come the entire time. You know, some people, you know, they can come the whole month, some people have families and stuff. They can. But we're like, oh, come for a week or come for however long you can. Right?
**A** (24:41):
Yeah.
**B** (24:42):
And I think that kind of stuff, even if along with being better for working in person, it also just gives you like the rapport that makes it much easier to work with people online after that because you know, now you're actually friends, not strangers.
**A** (24:55):
Yeah.
**B** (24:55):
Anymore. And so I think that's really important. And even just doing like mentorship in person meetings. Yeah. And then like doing like the continent wide stuff. Like I think part of the problem is like, okay, getting everyone in one place is like hard. Just, it's expensive. But it's also like jet lag and stuff like knocks out half your productivity. But it's like like, oh, okay, if you're in the U.S. why don't we have a bunch of people here meet more regularly or everyone in Europe meeting one city every two months or something like that.
**A** (25:27):
I hear you. That sounds great. Thank you so much for sharing your insights with other founders in our space. Appreciate you. Thank you for being here.