It’s too damn hard to build on Solana
Mert Mumtaz is on a mission to make building on Solana easier. The company he founded, Helius, enables this through affordable access to RPCs, blazing fast APIs, and webhook abstractions to help developers focus on their products rather than infrastructure complexities. I caught up with Mert to learn more about his views on building in crypto, and what he’s most excited to see as more developers enter the space.
EC: What are you trying to accomplish at Helius — both today and aspirationally?
MM: We started Helius because building on Solana is hard, expensive, and time consuming. Due to the sheer scale of the chain, there’s a lot of data processes and you need some high end hardware to get started. Its functional programming model is different and is also quite unusual for newer devs as opposed to something like Solidity, which is pretty much just Typescript. On Solana, state and code are decoupled, which leads to very interesting architectures and challenges. It’s also been around for less time when compared to other Layer 1s. When you add these up, it leads to a relatively high barrier of entry. People end up getting intimidated.
We created Helius to help people ship — faster, better and cheaper. We do that by providing access to the fastest RPCs at the most affordable prices. We offer enhanced Solana APIs, where we reverse-engineer and parse transactions to make them human readable, on top of other useful abstractions. We also have webhooks, which help with processing real-time data without having to set up advanced infrastructure like Geyser Plugins (which requires a good chunk of brainpower and hardware). The TLDR is that these things are resource-extensive and pose significant engineering challenges since Solana is essentially an insanely high-end computer that syncs at the speed of light. Our initial goal for at least the next year — and we’re not going to stop until we get there — is to make building on Solana an absolute delight.
Long term — once people stop worrying about infrastructure and these implementation details, once Solana gets to a state where people can have an idea for a product and they can just build without having to work with these low level abstractions — that’s when we’ll know we’ve made a dent. Then we can start expanding to other blockchains where people will have similar problems (though I’m sure Solana will keep us busy for a long time). But I’m not really a fan of how so many infra providers generally offer something very shallow and then duplicate it across chains. You really can’t empathize with the developers of that chain if you do that.
EC: Awesome. Given your depth of experience working in the low level infrastructure of Solana, what are some problems that you think people are not talking about but should be more conscious of when choosing to build on Solana?
MM: The biggest problem is that Solana developers don’t generally publish their IDLs (Interface Description Language). Stepping back a bit, in the Ethereum Virtual Machine, you have these things called ABI files. These files describe the inputs and outputs of a contract, as well as their types. On Solana, this ABI file, called an IDL, isn’t published often. In fact, only 6–7% of all programs have it. Moreover, most programs are closed source. This means when you go to a block explorer to look at transactions on-chain, all you see are randomly encoded strings. You have no idea what the data means (unless you know how it was encoded).
Helius reverse engineers these transactions to decode what they are. Instead of cryptic strings, you’ll see something like “this was a DAO vote on a Monke DAO proposal by Nom.sol,” for example. This is a very Solana specific problem and one of the main things we’re trying to solve.
Another challenge specific to Solana is that the hardware is much higher-end than any other chain. For example, our RPCs actually have at least 1 terabyte of RAM. Your Mac at home in comparison, usually has 8 to 16 gigabytes. As a result, people have to use Geyser streams on Solana, which help providers and validators get a real-time stream of new blocks, new account changes, and new transactions, amongst other creative solutions. This is quite hard to set up and work with as a newcomer, as you need your own RPC node and a good amount of lower-level programming knowledge. You’d need to learn Rust and potentially even develop plugins yourself, not to mention the complexities of managing bare-metal servers and general site-reliability challenges. We created webhooks to help abstract away all of this and make it easy for non-Solana devs to get started on the chain as quickly as possible with no background knowledge necessary.
EC: So as the complexity of blockchains gets abstracted away and building on top of Solana becomes a delightful experience, what do you think that enables or unlocks?
MM: We’ve essentially cheated on that question, as we don’t really know the answer. We simply build the lego bricks and people can build whatever lego structures they want.
There are some interesting use cases though — I’ve been talking to a large accounting firm that’s active in the space. Their clients don’t want to use Solana because it’s too hard to understand on-chain activity, and they’re worried about compliance risks. For example, if I perform this transaction and I don’t understand it, will it have a negative consequence later?
At the end of the day, it’s a numbers game. We want as many people to enter and explore this design space, similar to the early days of the internet. A lot of the cool things you see today, like Google, Facebook, Snapchat — these things weren’t built until decades after some of the foundational primitives of the internet were created. There were quite a few iterations and a lot of abstractions had to be built to enable more complex systems.
On Solana specifically, the goal is to synchronize information across as many machines as possible at the speed of light. That’s impactful for all kinds of markets, so that users can’t get arbitraged by some fund that may have an unfair advantage.
Generally, I think it’s still early and some of the ideas that will come out haven’t even been envisioned yet.
EC: What got you interested in the crypto space?
MM: I’ve been interested in crypto since school. I took a few courses in pure cryptography and information theory due to majoring in math and a good amount of my friends ended up at crypto companies like Element, Maple, Nansen and Flashbots. I’ve generally been around people that are much smarter who were building on crypto. I didn’t really take the plunge until two-and-a-half years ago, for two main reasons: one pragmatic and the other philosophical.
The pragmatic reason was that I was working at a fintech startup, and we were trying to fund entrepreneurs in Australia. I was in charge of the tech behind the fund movements. We had to jump through all of these different routes and jurisdictions to transfer money from Canada to Australia. I thought that was wildly inefficient — why couldn’t we just use USDC and send it on Ethereum? That would only take a few seconds. The answer I received was pretty much: “crypto is a scam, people use it for ransoms,” and I was pretty annoyed with the close-mindedness of that answer. A few weeks later, I started working for the team that handles money movements at Coinbase.
The more philosophical reason is that I believe that society as a whole is becoming too centralized. This was especially accelerated by the pandemic. I think that trying to combat the negative consequences of total centralization by building systems which enable decentralization is an undertaking worth taking. I also don’t think max decentralization is good where it’s total anarchy, but we do need solutions against total centralization.
EC: Outside of your time building Helius, what do you do for fun?
MM: I don’t know if this counts, but I’m pretty interested in DAOs and decentralized governance. I recently started a DAO on Solana called Lamport DAO. The goal is to create a community for Solana devs to lounge, talk tech, and accomplish things as a collective. It’s very early days, but it’s fun for me.
Also, I love cars, and any powerful machines really. I think there’s an interesting parallel with Solana, as it leverages the full power of very strong hardware.
EC: What’s in the garage these days?
MM: Right now, a BMW 3 Series, but I pre-ordered the new Lucid Air. It’s like a Tesla / Mercedes mix. I recently test drove one in Toronto and it was absolutely amazing.
EC: Awesome — so last one, how can folks get involved?
MM: If you’re a developer and want to build on Solana, you can look at our docs on the website at helius.xyz. If you want to find a cool place to chat with other Solana devs, you can follow Lamport DAO on Twitter. Feel free to shoot me a DM for an invite, as right now it’s invite only to make sure it’s a high signal-to-noise ratio. We will be opening up soon to the public.
EC: Thanks for talking with us Mert!
Notes
Elton Chang is an investor at Propel and the firm invested in Helius’ Seed round in 2022. Propel is an early stage venture fund investing in the new financial economy. For a full list of investments, please visit our portfolio page. If you’re building something in blockchain infrastructure, feel free to reach out to Elton (Twitter, Email)!
The dialogue has been edited for clarity and length.
None of the above is investment, business, legal, or tax advice and none of the financial information that might be contained has been verified or officially endorsed by Propel.
