From Dashboards to Data Primitives | Ben Stamm - Dune

All right, I'm going to start off today with a story. Nine months ago before I was at Dune, I was at my last job as a blockchain as a blockchain data engineer and myself and one of the one of the other engineers got tasked with this build an in-house Salana decoding function to extract the data from Meteor's token ahead of its launch. Track the data from Meteora ahead of its launch. so that we could be the only platform with Meteora data. So I was supporting the senior engineer and we went to work and we spent one to two cycles triing and erroring building this Python decoding function.

We got it working. We loaded it into Snowflake. We started testing it on blockchain data and we started testing on a day history. It all looked good. We started testing on a week history.

Okay, there's some bugs. We had to go on and and and fix fix those new nuances. Then we started testing on three months and a year and finally we got it working. This is back in March. At this point, we did not look at the snowflake bill.

We did not look at the costs. Fast forward 3 months later, Pump Fun is about to drop a token. So, we have an in-house salon decoding function. We're going to get the data before anybody else can. We're going to get it easily.

And so, we ran our function in Snowflake and running relatively intense compute functions in Snowflake. It's not cheap. And it and and what do you think happened when we ran it for the first time on Pum Funds contracts? It didn't work. Spoiler didn't work.

There was something unique about Pum Funds IDL. We had to go back in. We had to fix it and we had to test it on big data. and Salana has big data. And again, one to two maybe 3 weeks later and we got it working.

This time we checked the costs and uh then we huddled up to decide how to tell our boss about it. The whole time I'm sitting there and I'm wondering why would we not just buy this from Dune? And that's really the whole purpose for why Dune started is that yes, blockchain data is public, but it's hard to decode. It's hard to work with. It's fragmented across different chains with different virtual machines, which means different data architectures.

And it's constantly evolving. Even actually with the pump fund example, they the IDL changed one or two times like in the first two 3 months after launch. And so we had to go back and we had to modify our decoding function and run more compute etc. And again I think I think I already covered this. This is this is the problem that Dune was built to solve.

Dune has been in the blockchain data infrastructure business for eight years. You all probably have may have heard of Dune. probably have heard of Dune from the concept of dashboards. Dune is uh known of course for its dashboards, but I'm going to tell you today that in the past year, year and a half, Dune's launched uh so many new developer tools that dashboards are really only 10% of what Dune can offer to a developer to business from the context of using blockchain data to serve their business needs. And today I'm going to dive into the other 90%.

Dune moving forward will be currently is the only complete onchain data platform serverless onchain data platform. You take a look at the slide and there's really a lot of different things that you could do with Dune that we're going to dive into, but at the center is this platform that has full history for 130 blockchains that has inherent capabilities to server serverlessly. You know, you focus on your business. You can use the data and you don't have to focus on data when it's not your core business. So I'm going to walk through the components here, the big components that make up this data platform.

The first I already mentioned uh a little bit. It's the data store. So very different to most if not every other onchain data provider. If you want to work with data, you can work directly in Dune and you can access Dune's entire library of 130 chains, 60 million decoded smart contracts, curated tables with no access fee. You're going to pay for compute, you're going to pay on credits, you're going to pay for what you use.

Contrast that to Snowflake or to data bricks or to any other data infrastructure. You're now going to pay an access fee and pay for compute. And this is what makes this very important because it makes it's what makes Dune extremely unique in its positioning as uh a complete data platform. So we know about the charts but one thing that allows for this is that Dune's query engine is unique. It's a fork of Trino and it's optimized for blockchain data in a variety of different ways.

Um and that's all inhouse. That's all packaged as part of the data platform. You're that's what the users are using. And I'll go through a few uh I'll talk through two different ways where Dune's engine is customized for blockchain data and can do things that other data warehouses cannot. Um but you see the link at the bottom.

There's quite a few optimizations. So, you know, if you're really interested, you could go and you can take a look. But on the left, var binary data types makes working with hashes and addresses extremely easy. Um there's no you don't have to do any case matching. Um it's uh that's all handled by the query engine.

And on the right, Dune's query engine uh is compatible is compatible with embedded API calls and API authentication. So you could see on the left that Morpho is calling an API and they've authenticated their API on their settings and so they're calling the API getting the API output and funneling that into their query in one go. So went a little bit fast through the query engine because there's really a whole world of blockchain optimization there. Um but the next category you know the queries that are you can you can access Dun's data and queries from your application using a pretty large suite of data APIs and you know the most common use cases are to you know run queries programmatically or fetch query results but if you look on the left and you could see this on dune.com there are you know 10 different API endpoints that allow you to execute entire data pipelines.

They allow you to, you know, set up monitoring alerts on uh failed executions and and much more. Similarly, Dune offers real- time APIs, a set of currently there are eight endpoints uh real time under 200 millisecond latency that is an out-of-the-box solution to provide to power wallet applications. Um, one example of this is the DeFi endpoint which will return a user's DeFi positions across chains, across protocols into their wallet, whether that be lending, whether that be liquidity positions, etc. But this but this real-time API endpoint is also valuable for anybody who is looking to build with one of the select endpoints for uh a given use case as well. This is the real powerful slide that I was alluding to earlier on in the talk.

This is what separates Dune from the playing field. And Dune does have a data share business where we can data share data into your environment. If for whatever reason for security or for proprietary if you have high high volumes of proprietary data maybe you do want to pay extra and work in snowflake, bitquery, data bricks or another data warehouse. But fundamentally Dune can do what those data warehouses do without a data access fee. It's all pay for usage.

One of the big pain points that we see when we have these conversations with users who want to consume Dune data in Snowflake is that they have to lock into the chains and the data sets they want to use up front because from our standpoint every all the data sets that we share we have to pay egress we have to pay replication and whatnot. So they're going to say we want Salana, we want Ethereum, we want Optimism and we'll sign a contract and two months later maybe they want to work on uh Arbitum and now they have to go through the process of spinning up a new proc new contract and paying new money with Dune you know it's pay as you go use the entire data library get access to new data sets as they come on board and it's a much more seamless it's a better product if you're doing data analytics on blockchain specific data. There's no better place to do it than Dune, which already has the data. And finally, I will reference some of the connector tooling that allows users to use Dune from their own developer toolkit and developer environment. We recently launched uh a DBT connector which allows a developer to spin up a DBT repository, point it at the Dune data warehouse and again have access to the entire data library.

Um we have a SQL mesh connector, we have a TRO connector and really we're enabling anybody to run transformations on Dun's warehouse. So finally I'm going to leave with you guys with something exciting which doesn't currently exist but will be coming very very soon to Dune. Two things actually. So I had a video of the MCP on the slide but let me just say the Dune MCP is coming soon. DUNE's data available for AI agents available and optimized for AI agents.

And finally, the stable coin data set. This data set will be the best, most comprehensive stable coin data set that you can buy on the market. It will have stable coin transfers with labeling. Almost 60 to 70% of the addresses are labeled. 60 70% of the activities are labeled as well.

So, not only can you see who is transferring stable coins, but you could see what for, you could see if they're translate you're transferring stable coins to deposit liquidity into a a DEX or if they're borrowing or if they're peer-to-peer transferring. And this rich label thing is very uh you know I have to give a shout out to our engineering team because it took a lot of deep engineering into the traces into the logs and a lot of compute cost to actually seed this data set. So it's not that easy to replicate in addition to the proprietary labeling that we do have on our side that makes this data set a one of one. And lastly, I'll see as the time is running up, the stablecoin data set also comes with stablecoin balances, tagged and enriched balances, where anybody who's in the world of blockchain data knows how difficult it is to get a really good balances data set. And uh we're really excited about what we're going to bring to the market.

So feel free to come talk to me afterwards uh about any of the content that I talked about today. We have a booth and a fantastic team here and we're all happy to engage and uh thank you [applause]