S3:E4 Harnessing Blockchain Data with Jonathan Kau
On this episode of Archebyte we are joined by Jonathan Kau, Co-founder of Satsuma, a blockchain indexing and query platform.
We begin our conversation with Jonathan walking us through his founder journey and a 101 lesson on blockchain data. We discuss the unique challenges that web3 developers face and explore how Satsuma can ease some of those challenges.
We also touch on the challenges of selling to DAOs and navigating their unique decision making processes. We discuss the importance of giving customers a voice in shaping the product, while still maintaining a strong sense of direction and opinion about the solutions being offered.
📬 To keep up with the latest from Archebyte and receive articles and research from the rest of the Archetype team, subscribe to our newsletter: http://eepurl.com/iCApL2
- - - - - - - -
1:03 Founding Satsuma
3:12 Blockchain data 101
6:46 Challenges for web3 developers
8:07 EVM compatibility
10:38 How Satsuma works
11:22 Centralized vs. decentralized data providers
14:57 Building Satsuma in a bear market
16:57 Satsuma’s building strategy
19:48 Doing business with DAOs
23:57 Jonathan’s advice for aspiring founders
👋 FOLLOW US
Hello everyone and welcome back to Archebyte. I'm your host Katherine Wu.
Each week at Archebyte, we have on the smartest voices in the crypto industry to tell us about what's top of mind for them. I'm joined today by Jonathan Kau, Co-founder and CEO of Satsuma, formerly Jonathan was a product engineer at Y Combinator, Founding engineer at Penny and Co-founder and CTO at 17cow.
Welcome to the show, Jonathan.
Thanks, Katherine. Great to be here.
Just for our listeners out here, this is Jonathan's first ever podcast interview, I'm sure he's going to do amazing. So Jonathan, tell us about your founder journey. What led you to start Satsuma and also what is Satsuma?
Yeah, great question to start us off. So the story really starts way back. I got into crypto back in 2014 actually. I started building an app on Bitcoin back then, and 2016 / 2017 came around. I'm an engineer by background, so I started tinkering with Ethereum and playing around with NFTs. And it really wasn't until last year, early 2022, when I started to notice that there are a lot more legitimate engineering teams that were building wide scale dapps on blockchains.
And what really happened was my co-founder and I had started down a completely different direction as many founding stories go. We were focused on product analytics for dapps and web3 companies, and as we were building our own data stack, we realized that a lot of the tools that we had at our disposal were pretty hard to use.
And that's what led us down the path of working on Satsuma, which is a blockchain data platform for developers to generate custom APIs off of their own smart contracts.
So basically you're building a very important crypto infra company. And, you know, admittedly, sometimes when you think about an infra company or infrastructure, it's not the most exciting topic. That doesn't mean that what you’re building isn’t important, but It's not like a sexy topic about tokens or token economics. However, I do think in a slower market the folks building infra really get to shine because this is when a lot of the newcomers sort of leave and then all the problems that came from like the last hype cycle kind of really are showing.
Right and then for folks like yourself who built infra I think that this is a really good time to do it. And Satsuma, specifically, we're kind of venturing into the blockchain data space, but for folks who aren't as familiar. Walk us through basically everything that plays a part in blockchain data. Like basically give us a walk through.
Yeah, I like to start from the most basic level and forgive me if you're listening to this and you're an engineer who understands everything. But where I start is really with what underpins a blockchain, and that's just a bunch of nodes. And you've probably heard of a blockchain node before they perform a lot of tasks at the most basic level.
They have a database somewhere that stores a bunch of data like blocks, transactions, events that are happening. And then on top of that, they have this consensus mechanism where they're talking to each other, exchanging data and so on. Now, if you go one step higher, I also want to define another term which you probably heard a lot. RPCs or RPC providers.
What RPC really means is it's this generic term for an API interface. And what happened was Etherium popularized what's called their standard EVM compatible JSON, RPC interface. And because subsequent chains all adopted this interface, we now just call that interface RPCs. And RPC providers also provide that same standard interface. And now we think of Alchemy, Infura and Quicknode as RPCs and RPC providers.
So we're starting with nodes. They store a bunch of data. Now in terms of extracting that data, what you'll find is that the JSON RPC interface that a node exposes is pretty basic. It lets you get things like, Hey, give me this one block or give me this set of transactions. But if I want to answer, what's the TVL of Uniswap’s USDC / ETH Pool at this time is that's pretty hard to answer.
And so what you need to do on top of that low level data is called indexing. And what indexing really means is you pull out a bunch of raw data, you do some preprocessing, and that preprocessing makes it easier to then at query time, answer sorting questions. And finally, for the last piece that I'll quickly talk about here, where sub graphs and the graph comes in.
So the graph came along in 2018 and they've built some amazing technology to help developers in the ecosystem with this problem of indexing. And the main use cases here are that they allow you to store less data onchain, which is very expensive. And as I mentioned before, you can preprocess it. So the graph came along in 2018.
They provide a really easy interface for developers to specify what API they want to generate out of their smart contract data. And it speeds up your development time dramatically. Instead of spending a month, maybe now you spend a week on your data. That's the high level.
I love getting my guests on and then just making them actually give me a history lesson instead. Okay. So you mentioned subgraphs. You kind of alluded to challenges with querying. Can you expand on that? So what are some of the unique challenges that web3 data engineers have to deal with that aren't part of the, I guess, like web2 world?
Yeah, yeah. Great question. So in the web2 world, all of your data is in your own centralized, siloed database somewhere. Maybe you have a progress database living in your cloud instance. So this is what Facebook does. For example, they just have a bunch of siloed user data. Now, if you talk about web3, a lot of these projects are storing their data on a shared database.
They're storing that data on blockchain nodes and so you have this extra step of extracting and then indexing that data. This can often add up to, say, 30%. Sometimes I've even seen it take 50% of your development time. Just focused on this piece of extracting and then processing the data. And it's not something that you really have to handle in web2.
As I mentioned before, the interface for interacting with blockchain nodes is pretty basic. It's a standard that every single chain uses. So you're not going to have fancy things like, give me a Uniswap TVL in that. And so you have to take the really basic elements and build up from that data yourself.
I want to ask about how this process works, I guess, across different L1s.
Great question. Depending on the type of L1. I'll generally break this up into, say, EVM compatible L1s and non-EVM compatible L1s. When you hear the term EVM compatible, one of the main pieces is that for a developer, you get to use the same exact JSON interface. So if I'm a developer building on, say, Celo, which does have EVM compatibility, I don't have to change that much code on my front end or back end in order to say, get a transaction or get a block.
Now, if you're talking about other L1, say Solana, they do have completely different interfaces. And that's why it is a much bigger challenge for a dapp, say a protocol to onboard Solana versus, say, another EVM compatible chain.
And I think the fact that Ethereum as an ecosystem has an L1 has been around for longer, meaning that there's more info around it that makes the experience easier. So then back to Satsuma. What role do you play in the blockchain data maze and how does it work?
Yeah. So what we started hearing from Dapps and projects was that the graph is an amazing piece of technology, but there are some kinks in using it and there were a ton of projects that had written their subgroups and have invested a lot in this indexing paradigm, but were facing issues like performance. So how long it takes to index your data, how long it takes for data to show up and also reliability.
So how is my API actually serving me data at all times? And so where we come in is that we have a complete drop in replacement to the graph. So it takes you less than 5 minutes to switch over and we provide you with a lot of the benefits that I mentioned before. We improve indexing speeds by 2X or higher and we have a ton of developer tooling that saves you time on your iteration.
So how do you 2X like the speed of indexing data? Sounds like too good to be true a little bit. So I just like how?
Yeah, yeah. We get this question a lot and I can't give away all our secret sauce, but we have gone really deep in looking at how some graphs work. We have optimizations on the inference side anywhere from nodes, the database to whatnot. And a lot of it also involves pre-processing, a lot of data and extracting that ourselves.
Maybe in the future, instead of a sales call, you just send them this podcast like Katherine asked everything you're asking me. So just listen to this.
Yeah, yeah, it can go into our FAQ.
Yes. Well, we can't talk about anything in crypto without bringing in the topic of decentralization. And so as you think about building Satsuma in web3 decentralized worlds, how do you think about the difference between building like a centralized or decentralized data provider?
Yeah, that's a great question. Comes up a lot as well. So just to be clear, we are a centralized data provider. We sell typical SaaS or software offering. And I think what I believe is that you don't need decentralization in every single part of your life or in terms of a business, every single area. What I think about is the term sufficient decentralization.
So you have decentralization where it really, really matters and you don't need it for the places where it doesn't matter. So where it comes to us, what does matter with a data provider is a few things. One, you want to make sure that they're serving you the accurate data. So that's pretty straightforward. You want that data to be reliable, and ideally you want to reduce vendor lock in more than anything else.
And so in terms of our customers, what we found is that they actually don't care that we are a centralized provider because we're all following the same open source subgroup specification. It's kind of similar to how with the JSON RPC interface that I described earlier, for nodes, they're all using the same interface. So you can swap between pure or Quicknode or Alchemy pretty easily, and you're never locked into one of those.
We think of it similarly with some graphs as well. So a customer can switch to us. They can switch back to running their own infrastructure if they want to. And it's not quite the centralization from the data provider point of view, but it's the centralization because you aren't locked in to one provider. And we found that many of our customers really, really enjoy having a point of contact to go to for technical support whenever they have issues.
Yeah, I mean, one thing I always say a lot with friends and family about crypto is that crypto is ultimately about choice. The difference between having a choice and not having choice, at least where crypto comes in, is, you know, if you're born in the U.S., you're not opting into anything. You're born, you use U.S. dollars, you have a U.S. bank account.
Like all of that. Right. There's no like, oh, when I turn 18, let me opt in to something. It's just. It's just there. Right? So, like, when I phrase it like that, where I'm like, I'm not trying to convert anybody in crypto. I'm just letting you know that like, it is a choice you can make to opt into a whole system, right?
And like I started off as a whole new financial system, but now it goes all beyond that, right? Like, sure, you can choose to use all these centralized apps, but you can also opt in to any, you know, web3 social or, you know, like all that stuff. So vendor lock in is like a subset of that.
Yeah, for sure. For sure. I definitely agree with that.
I'm very curious. You know, since we made that joke about sending this to any of your future customer calls, who are these types of customers? And how does building in this current market affect your strategy and how you're positioning yourselves today or tomorrow or, you know, in the future?
Yeah, yeah, definitely. So to answer the first one, just on our customers, we are supporting some of the top users of subgraphs, and just top projects in the ecosystem overall. So we're supporting teams at Decentraland, Aragon and we're working with Treasure and gaming. These are all projects, doing great work. In terms of how we think about building in a bear market, it allows you to really focus and have clarity around what you're doing.
I think it's actually a surprising positive for us because you remove all the distractions of trying to go to all the events and trying to form partnerships with every single person possible. What you try to focus on is how can I serve my customers and what do they actually want for us? My co-founder and I, we're pretty frugal by nature already, so it doesn't affect us too much.
But because of the bear market, we try to stay lean as a team and we're trying to grow sustainably off of revenue alone. And whether in a bear market or in a bull market, I think if you remain focused on what customers actually want, then you're going to do well. Another thing that I'll add is, to be honest, many people ask us to this question about building in a bear market, but I almost didn't realize at some point that we were building in a bear market because we were just in this whirlwind of building what our customers wanted, working with our customers very closely and growing with them.
And I just looked up and people were asking me every single time I had a call with them, Hey, what's it like with crypto being down all the time? So that's from my perspective as an infra provider.
Yeah, yeah. No, that makes sense. If I have any founders listening to this, you know, I'm sure this is like a process every startup goes to, right where you're like, okay, let's identify our potential users, right? And you do this exercise and then, yeah, but, but you're a little bit further along like you have paying customers. And so how do you go about basically figuring out especially now that crypto is so big, like what sectors to go after.
I think you mentioned gaming just now. I don't know if that's like a core focus. Like how do you go about, I guess like mapping that out, prioritizing which sectors or is it just kind of like, let's try it all, see what sticks?
Yeah, to be honest, I don't have a great answer because everyone's founder journey is different. Some people focus very deeply on a sector that they know a lot about. Some people do this sort of market analysis type motion. I will say it really goes back to the saying: customers will pull the product out of you if there is a product to be built.
We started out and I think we tried to be too opinionated over what customers wanted, and that actually backfired in some ways. People were telling us, Hey, yeah, I actually had this problem over here that you should pay attention to. And it took a little bit for us to finally pay attention to that problem and switch over. I think there are always exceptions to every rule, but generally for most founders, if you're in this position, that's what you should focus on.
It is a delicate balance between building the products that you set out to build and getting sidetracked by a million different requests. And I do think that it's a balance. Yeah.
Yeah, yeah. The way I describe it is you should allow your customers to be really opinionated about their problems, but you should be opinionated about the solutions themselves and make sure that you delineate between those two and try not to be opinionated about their problems and vice versa. Don't let them shape the actual solutions.
Wise words. Yeah. There are a lot of conversations being had right now within my circles about just why are DAOs so messy and prone to mismanagement? And we need a CEO for a DAO. And I do think part of it is like in the case of a DAO, the customers or the token holders’ voices are really, really loud.
And right now, like even stewarts of a DAO, whatever you want to call them, elected contributors of the DAO. I think it's lost in this balancing act a little bit. Right. Like as a founder of a web2 company, I think you have a little bit more say in the direction, but for a true like web3 DAO I think we're struggling with that balancing act right now.
Yeah, yeah. We work with many DAOs and decentralized organizations. The types of organizations that I think do feel like they're working very well are when you have a core group that are making a lot of the decisions, maybe it's not one person. It's almost like a council in some ways or core contributors. But you still have people who are in charge of many of the decisions because it's really hard to make decisions, especially small ones, as a collective.
Yeah. Oh, my God. Actually, what is it like selling to a DAO? What is it like as a potential service provider to a DAO?
Yeah, I'd say it helps that I have been in crypto for a little while and I'm an engineer, so I feel like I have an advantage on how to speak to these people and go to where they are. It's definitely different from sales in the traditional software world. You can't just cold call people or hit them up on LinkedIn.
Oftentimes I'm on Telegram and Discord Channels, completely different ways of getting in touch, and then the actual sales process for a DAO is really tricky because you have to figure out and navigate whatever process that they might have in place for looking at software. There's trade-offs. On the plus side, if we find our champion, then they're going to have a lot of sway and they're going to be able to really help us out and help us navigate.
But the downside is a lot of these processes are not built and it's just going to take a lot more time than you think.
So just to sum up, what you said first is harder to identify who to talk to in a DAO, whereas in a normal company to kind of figure out who you are, who you're selling to is it hard to have sales calls that are open to everybody? Like, have you had to do a sales call where it's like 20 random people online and you like, don't know talking to have you had to do that?
Yeah, not exactly that. Typically, we will track down the right person and that's typically an engineer at the end of the day, whoever is going to be using our product. And what happens is we have to kind of backtrack from, hey, this user really wants our product to then figure out, okay, just like in a normal organization, we just have to find our way to navigate a DAO in order to get to the sale.
I think you have interesting, weird challenges, like is there going to be a legal entity to sign the documents that often comes up? Are they going to pay in crypto, for example? And so on.
Yeah, I think as more service providers start to engage with DAOs and sell to DAOs and as the landscape matures. Like there is DAO native tooling that makes sense. But there's also the need for just like your, you know, web2 software model and I've never actually had a conversation with someone who runs a SaaS business selling to a DAO because I feel like that's just such a different way of working.
Like, I don't know if you end up having to hire salespeople who are like crypto native or something, but yeah, you have that. Yeah, but that seems like a whole new challenge.
Yeah, it's totally different. This is a sidetrack. But I was at ETHDenver this year and I was meeting up with one of our customers, but I had never seen their face before. I only knew their pseudonym. And so I show up at one of their happy hours and I'm like, Hey, do you know where so-and-so is? And eventually I tracked down the right person.
But yeah, I had never seen them before.
Yeah, well, so I come from this, like on the other side. So, like, I am an elected steward of the ENS DAO. And like, literally this morning we had kind of a potential service provider who wants to build something for our proposals come in. And as he was presenting the demo, I was like, you know, I don't think he knows anybody on this call right now.
There's like 20 of us. No one has a camera on like, you know, maybe like he knows me, but like he's presenting to a room where he doesn't know who they are. He doesn't know what they do, or he doesn't know how they're involved in the DAO. It was just an open call and he basically was like, you know, showing the demo of what this thing would look like.
And you just gotta roll with it. You should write “how to sell to DAOs.”
Yeah, that's a great idea. I'll que that up for our blog.
Yeah, I think this is a good sidetrack to bring it back. I have one last question for you, which is, do you have any advice to any listeners out there who would like to start a company one day?
Yeah, that's a really healthy question. I'd say give yourself time because this is a really, really hard process. And when you see success stories, they're always filtered through a specific lens and narrative that they want you to believe and the journey is really never that straightforward. Almost all successful companies have had ups and downs, so it's really going to take a lot more time than you think.
So I think that's one. Another one is really pretty basic. Get a therapist. You're going to go through the strongest emotions, both on the positive and negative that you've ever experienced as a founder. And I thought I was a well-balanced person before I started a company. And I've learned a ton by having someone who is professionally capable of helping me with this stuff.
That's actually great advice. And I think something that should be honestly talked about more so, not as obvious as you think.
Yeah. Yeah, for sure.
Awesome. Well, thank you, John, for taking the time to talk to us about subgraphs, RPCs, building in a bear market, crypto infra, selling to DAOs, and advice for founders! We packed a lot in 30 minutes. But thank you for your time and hope to see you soon.
Thanks, Katherine. This was great.