Data, AI and the blockchain paradigm shift with Chainbase COO Chris Feng

[00:00:03] Chris Feng: We are building the very first data infrastructure for the whole industry. And he made 30 times in this investment in about a week. It feels like every company has shifted into being an AI company, right? Every blockchain into an isolated island. Oh, entrepreneurship is always with challenges.

[00:00:23] Elisha Owusu Akyaw: Hello everyone. Welcome to this episode of Hashing It Out with your boy, @ghcryptoguy. Hashing It Out is a podcast distributed by Cointelegraph, powered by Cointelegraph, and today, we are going to be talking about a topic that people don’t probably usually look at. I’ve seen some people talk about it, but it’s not a conversation that a lot of people have. It’s about the role of data in the blockchain space and how the blockchain space is built on data in and of itself. We put data on the blockchain all the time. One of the reasons why everybody loves the blockchain is that it’s easier for us to follow, track and verify data. Joining me for this conversation is Chris Fang, who is the co-founder and chief operating officer at Chainbase. Hello, Chris, and welcome.

[00:01:11] Chris Feng: Hey, Elisha. Thanks for having me here.

[00:01:13] Elisha Owusu Akyaw: Amazing. So, we are going to go to the boring questions first before we get started. Can you kindly tell my audience who you are and how you got into the crypto space and what you are doing at Chainbase?

[00:01:26] Chris Feng: Sure, let me start my journey. Since 2016, I was working as a consultant at Bain, which is the management consulting firm worldwide, and found myself in Tel Aviv for the cross-border M&A deal. While I was there, I cannot help but notice this vibrant buzz around crypto, and even we call it a blockchain at that time. Everyone was talking about it, and I just took the price of the blockchain. At that time, it was around 600 USD, if I remember clearly. Honestly, I have no clue what it is. I didn’t know why I want to buy it or how even go about it. I just know it’s a big thing. So, fast forward to 2019. I had a dinner with Justin Sun, the founder of Tron. He was launching a coin called BTT for the BitTorrent acquisition, and it was going to be launched on Binance Launchpad. That was my first time to hear launchpad as well. I remember that sitting there, I’m still questioning why do we need the coin and how can we capture value through it. And what is the key driver to pop up the price? I have no idea. But one of my friend at the table, who is now the founder of Bitcoin layer-2 project as well, he decided to buy it, and he made 30 times in his investment in about a week. I thought, wow, this is crypto. That moment really opened up my eyes for the whole blockchain space.

After that, I start diving into trading crypto on secondary market, and eventually, the concept of Web3 came onto my radar because I have an experience in both Web1 and Web2. So, I realized that at the beginning of the new technical cycle, the best way to tackle that is to jump into as an entrepreneur, not just investment or trading. So, at the beginning of 2022, I decided to co-found Chainbase with my partner, and our vision was very clear from that time. We seek onchain data as the paradigm shift of the whole data category, and onchain data probably not only data but also new asset class. And more importantly, that while blockchain is the groundbreaking shared ledger, it is more importantly to see it as an incredibly data-driven segment. The role of data in a category is being more important than whatsoever before, and we are seeing the paradigm shift. So, we dive into it, and we deal with onchain data for more than two years. And when I reflect on my journey, I feel like everything just lined up perfectly. If I had jumped into the training at that time, probably I would not be an entrepreneur and sitting here to talk with you. But in any way, those moments are unforgettable. And I’d love to dive into data with you guys for the coming hours.

[00:04:16] Elisha Owusu Akyaw: Sounds like an interesting start to your blockchain, Web3 journey. So, let’s talk about Chainbase. What is Chainbase, and what are you all building at Chainbase?

[00:04:25] Chris Feng: Yeah, the core approach of Chainbase is that we are building the very first data infrastructure for the whole industry. As I said previously, what we see, Web3 is like a paradigm shift for data. And how Chainbase tackle this problem is that we built the crypto fundamental layer for anyone to interact with our data. Then we just divided all these things into 14 new consensus architecture to make sure everyone can rollup data into our network, which can be rewarded by our own token. And in our network, we’re going to process all this data into a high-quality format based on transparency and process of it. Then, we have the high-quality data based on the standardized schema that we designed for the industry, which we call the manuscript. And based on the manuscript, all the developer can generate customized API or predefined API and even crypto model based on the model capability building in our network. So, in this way, we want to unleash all the value of data, and we want to create a bunch of standardized, high-quality data for not only developer to interact with, but also for AI model to training. This, we believe, will be the brand new data stack that optimize all the value of data. Because in our mind, our data is not only the data category but also new asset class, which allow us to provide data assets for the whole industry, then monetizes it to give you an idea on our impact. During the past two years, we’ve attracted over 15,000 active developers for partnership with 8,000 projects, parsed the 20,000 smart contract, and generated 200,000 data sets, executed nearly 500 billion data calls in total, the stored data at the database level, and collaborate with more than 20 public chains. The number goes up to 200 chains by the end of the year, covering all major EVM and non-EVM ecosystem as well as leading public chains.

[00:06:31] Elisha Owusu Akyaw: That sounds interesting. So, let’s break it down this way. We have some data aggregators like CoinMarketCap and CoinGecko for the average crypto user. What is Chainbase doing that these platforms aren’t?

[00:06:47] Chris Feng: You mean for end-user, right?

[00:06:48] Elisha Owusu Akyaw: Yeah, for the end-user.

[00:06:49] Chris Feng: To keep it simple because I don’t want to dive into the technical details because, honestly, most users probably are not so interested in that. So, we are creating a brand new data stack that’s going to empower crypto AI in total. So, the great thing about it is that any data on a blockchain gets integrated into our network, and everyone who contributes data is rewarded. We also have this cool feature where we convert raw data into high-quality data. This transformation involves a lot of developers and community contributors. Our role as a platform is to provide a standard tools and environment, while the contributor who helped with this process can get shared in the network’s reward and also tokens. So, in a nutshell, our goal is to activate the entire data intelligence ecosystem with just one token. We’ve already launched entry points for developers and contributors of our testnet, and we would love everyone to jump in and participate.

[00:07:53] Elisha Owusu Akyaw: Sounds great. So, I think the next interesting thing that I want us to touch on is the conversation of AI and blockchain technology. We’ve seen a lot of projects building at the intersection of AI and Web3. What’s your take on how AI can boost the Web3 space, and how is Chainbase integrating AI?

[00:08:13] Chris Feng: For sure. There is no doubt that since GPT came onto the space, it feels like every company, big or small, no matter where they’re located, has shifted into being an AI company, right? Or at least a company powered by AI. It’s like the new wave of AI technology is putting everything in its wake. Remember the internet boom, right? Is this similar, like what we’re facing right now. So, every company has to have their own website and the application marketplace or even membership system over that time. So now, if we zoom in on crypto, crypto AI is definitely the hottest topic right now. It’s everywhere we look, in discussion at events, you name it. But we think that there are different approaches. We can really break it down into two categories. The one is crypto for AI. The other is AI for crypto. The key difference is what problem you are aiming to solve. If you are tackling issues with AI and using crypto as a tool, that fall under crypto for AI. Think about the numerous computing power platform and data labeling platform, they fit right into this category. They utilize the unique feature of crypto like incentivize, network collaboration, security, and decentralization to improve AI workflow. It’s fascinating space to watch, right? But when you look at the AI and big data segments on CoinMarketCap, you will notice there are not many mature project listed. The top ones you will find are Render, Bittensor, [inaudible] and a few others. It’s clear that every company is eager to jump into crypto and AI intersection. But what are the real-world applications, and how can we truly empower AI on a larger scale with the crypto field, or vice versa? It’s relatively easy to wrap the natural language indexing or interface around the existing business or quickly utilize GPT for prompt engineers, but I believe this approach doesn’t align with what is native crypto AI project should embody. 

So, how do we connect as the native crypto startup with AI? Fundamentally, we believe that among the three essential elements of AI, data computing, and model, data plays the vital role. Data processing where all AI journey begins. While some may say that data is the basic element with low entry barrier, but the truth is that for top-tier model, they all rely heavily on high-quality data input. For example, GPT-2 was trained on, if I remember clearly, 40 gigabyte of human-filtered data. While GPT-3 was trained on 500-something gigabyte of human-filtered data, by 40 terabyte of raw data. The larger and better of data set, the more likely the model is to excel. So, Chainbase, as one of the leading omnichain data network with over two years of delivering real data utility, we’ve built a solid foundation for creating crypto foundation model. What is particularly fascinating is that the data on our network is in a standardized format that is AI-trainable, making model training and usage much smoother. Many projects over the past two years have leveraged the data from teammates’ data network for model training, including ecosystem project like Scopechat, Hype AI, Ponder AI, the Agent.AI and more. So, our current goal is to shift from serving individual projects to providing essential building AI capability for the entire industry.

[00:11:54] Elisha Owusu Akyaw: I think one of the things that we’ve had in the crypto space over the years is every time there’s some sort of trend, people simply jump on the trend to seem relevant, or at least like attract people into their project. Beyond what Chainbase is doing, what’s your opinion on the general application of AI in the blockchain space, and where do you think we are heading towards in terms of blockchain AI application in the blockchain industry?

[00:12:23] Chris Feng: Totally. I believe that all the technical journey, I mean, the new technical journey, is going to be head up or piped up by application for sure. All the infrastructure are supporting for the application. But the tricky part is that up to now, the majority of the value is captured by infrastructure, as we can see that, for example, for the asset issuing and for gas fee generation, All this goes back to Ethereum, not the application on top of it. And right now, the application just play as the traffic generating or the user onboarding kind of role. So that is not the way what we like for the whole industry. That’s also the reason why we want to tackle data and also building the foundation model for the whole industry. We want to lower the entry barrier. We want to further drive the whole industry growth. In what way? Integrate all the onchain data because there is a project who want to build on Web3. They cannot get rid of data because all the Web3 project are based on data, right? So we want to tackle this problem, integrate all this data, and provide a lower entry barrier for applications to use data, to track data, and even to monitor all the things happening onchain. Then, the model will help them to lower the barrier to analyze the value of data. For example, if there is fully onchain game, all the gaming mechanism, scenario, transaction will be happening onchain. Now, how can the project to offer a smoother user experience for their users? They need to deal with data, but we want to handle all this. They only need to care about the user experience, not all their tokens and the gaming mechanism. So, this is the way we help the industry and will further drive the growth.

[00:14:14] Elisha Owusu Akyaw: So another important aspect of growth of platforms like Chainbase is developers. We’ve seen a lot of Web3 projects take on various approaches in their attempts to attract developers. What’s the strategy for Chainbase, and what has your experience been trying to get developers to build with Chainbase?

[00:14:36] Chris Feng: Yeah, this is the good question. We always been developer-centric product because we know that our core users are developers, and the company is also designed for developers. You know that the 80% of our employees are developer background. That give us more understanding on how developers are thinking about and how we can tackle them. We have a tremendous respect for everyone in the crypto data space, and we’ve built solid relationships with all of the developers. Speaking of growth, as I said, we are a developer-first company from day one. Our main focus has been how to serve developers better because they are truly our core user. We always listen to their feedback and aim to provide the most generous free plan we can offer. That’s why, over the past years, all our interfaces have been designed to be super developer-friendly. In fact, this did create a bit of a barrier for end-user or retail user to use it, but we are working on it. This year, we rapidly rolled up the building features like AI model, natural language indexing to make it easier for both developer and retail user to jump into and start using our platform.

[00:15:51] Elisha Owusu Akyaw: Interesting. So, let’s dive into some deeper conversation about Chainbase. You’ve seen some interesting growth, surpassing 500 billion data calls with over 15,000 developers. What would you attribute to this growth? It looked like a bull run at the start of the year, but now it looks like we’re in a bear run. So yeah, what would you say has led to this growth?

[00:16:14] Chris Feng: Yeah, that’s a good question. The core of our approach is to always listen closely to the voice and feedback of our key users. As I said previously, unlike many crypto projects that simply need to find a fit for their token and narrative, we recognize that as the infrastructure provider, our responsibility is to deliver a universally applicable capabilities, just like cloud services. This means our priority is to serve immediate pain points or high-frequency demands. More importantly, to deeply understand and cater to the fundamental demands for project and developers. They are the backbone of the use case. We really don’t believe in creating a bunch of feature or product just for the sake of it, especially if no one uses it. What we are passionate about is finding real-world scenarios that a lot of people are engaging with and then using our product to meet their needs in the best way possible. That’s why, as the infrastructure provider, we’ve achieved eight times growth in our developer community during 2023. We’ve also had a lot of public chains reaching out to us, looking for our help in onboarding new developers. That is an amazing time for us. Another thing I want to highlight is that building on this strong foundation, we’ve recently opened up the capability of our network to the wider community as a first step to decentralize the whole data network. Our hope is to encourage broader participation in the construction of a truly data-centric consensus. We warmly invite everyone to come and experience the new frontier.

[00:17:52] Elisha Owusu Akyaw: Sounds interesting. So, I think the other problem that we face in the blockchain space is the fragmentation of data and the fact that we have multiple blockchains using different, should I say, code bases, or just like a different approach to how a blockchain should be built. So, we have multiple layer 1s, layer 2s, and lately, we are seeing layer 3s. How are you trying to solve the issue of that fragmentation and making sure that it’s easier for people to track data across multiple blockchain networks in the multichain world that we find ourselves in?

[00:18:28] Chris Feng: This is also a good question. That’s why we call ourselves the largest omnichain network. I think we are the very first company to tackle omnichain or chain abstraction question from the data perspective, because as you said, there are many chains has been built up recently, and we believe there will be more and more. It makes that every blockchain into an isolated island. And how to integrate them together into the continent, this is the key because, as I said, no matter developer or a project, they don’t care about the single data. They want aggregated data based on a business sense based on the business knowledge. That’s why we designed the omnichain network, because we see all the data onchain as the whole thing, as the whole pie, and we just integrate all the data and define them in a business view like we’re going to abstract the data based on wallet, based on asset and based on different features. Then, give the end-user more idea of this type of data is solving this type of problem and whatsoever. So, integrate everything is one key that we need to dealing with. And the other thing is to how to offer compatibility for all this data. If we position ourselves as the universally applicable data layer, which means everyone can leverage the data capability offered by our platform to their project. We want everyone to take us as a data back end so that they don’t need to deal with isolated data in each chain. They can use the whole pie together based on our offerings.

[00:20:08] Elisha Owusu Akyaw: I’ve seen some talk on your social media about the crypto world model. What does that mean for the space?

[00:20:14] Chris Feng: Yeah, the crypto world model is innovative thing that we are doing right now. We have always strived to elevate the intelligence level onchain. Transforming high-quality data into intelligence requires us to build the crypto foundation model, or even we call it a crypto world model. This model, purpose-designed for crypto, learns from all onchain data, knowledge and information. It boosts the high performance similar to model like GPT while being capable of translating hidden crypto insights into transparent and traceable pieces with robust security and trustworthiness, and all onchain data used for model training is sourced from the network’s processing data, along with various offchain data sources. This model aims to translate massive onchain dark information into comprehensive intelligence for humans to use it. Our model training extract knowledge from both language model and the Chainbase omnichain data network. It uses the decentralized DORA, which is the purpose-designed algorithm that we use for the model training. It is to break down a large language model into the magnitude and the direction matrix. Those models can get together to generate a parameter matrix outlining all possibilities and decentralized training, insurance data confidentiality, and model robustness and result transparency for each result. Our foundation model provides a comprehensive reasoning onchain, producing transparent and trustworthy results. It learns crypto patterns from the vast online and offline data, as well as causal reasoning. This model comprises a powerful LLM paradigm matrix that rules along the crypto-native paradigm matrix. We also use RAG to connect the model to the real-time, onchain data network and internet eye reader in order to deliver up-to-date intelligence with readable argument. Users can create task model as well to tackle their specific business challenges.

[00:22:18] Elisha Owusu Akyaw: Talking about challenges, what are some of the challenges you faced while building Chainbase?

[00:22:23] Chris Feng: Oh, entrepreneurship is always with challenges. I think the main challenge we’re facing right now is how to generate data-centric consensus and how to spread the word out globally. Because as you may know, there are a lot of data indexing products or data analytics products tackle the portion of the challenges, but we want to do the problem in a more fundamental way. That gives us more challenges to pursuing everyone to leverage our standard to deal with onchain data, right? We deal with all this during the past two years by our own force. Also, we have a lot of data, community and data scientist work with us, but if we want to enlarge this data-centric consensus, we need to put more effort on end-user training and developer training and also the workshop. More incentive, other steps. Yeah, but we are excited to provide the universally applicable data layer to serve the all industry.

[00:23:22] Elisha Owusu Akyaw: Considering the current crypto landscape and the applications that we’ve seen come into this space. What do you think the future of data aggregation looks like? And what are the use cases of platforms like Chainbase that you envision in the coming months that would shape or change the crypto landscape?

[00:23:43] Chris Feng: Yeah, I think there are two examples that I want to answer. Your question, for one side, is what kind of data? I mean, new category of data will be generated or will be rolled up onchain because, as we all know, that on a Web2, along with the native data and the cloud computing technology, there will be more and more data generated on cloud, which makes all these internet forms up. But right now, on the Web3, the major category of data are related to transaction one. In the future, there will be more behavioral data or personal characteristics onchain that will give us more room to build innovative applications based on data. That’s what we are building right now, because the reason why we design an open gateway for anyone to provide data to our network, then we offer incentivize, is that we encourage everyone to take onchain data as the new data cloud or the new database to secure their data or store their data. In that way, we’re going to have more resources to work with. Then we’re going to cook on some juicy results on it. There’s one thing. The other thing is that I want to see some new trading platform that will fully leveraged by onchain intelligence because, as we know that previously, all the information is on a knowledge base. It’s not an intelligence base. All the trading is based on what smart contract did and how to monitor smart contract. But what if we feed in all these smart contract labels and all these parameters into the model, which is similar to what we design right now for ourselves, then I believe there are going to be some new intelligence generated. We don’t even know how can it generate it, but we can leverage the intelligence and treat the tokens, either onchain token or offchain token on a CEX, in a better way and generate more profit based on it.

[00:25:45] Elisha Owusu Akyaw: So we’ve discussed what’s next for data aggregators in the Web3 space, but what’s next for Chainbase specifically?

[00:25:54] Chris Feng: Yeah, we’re about to TG by the end of the year. And before that, I think we still have two important things to achieve. One is that, as I said, we need to onboarding as many developer project managers as possible because we believe the platform we design is going to benefit everyone, once you have onchain data needs. So we have a lot of user onboarding campaign and we have a bunch of developer incentive program, and we invite everyone to join us and help us to process all this, entering data into a standardized and high-quality format so that everyone can enjoy the benefits of data. This is one thing for us, very important. The other thing is that, as I said, we are training the crypto foundation model, and we have 200 million crypto native parameters. And we also leverage the open-source large language model. I believe we are going to open-source 70.2 billion parameters model on [inaudible] by the end of this year, and this open-source crypto-native model gonna even bring down chain to the internet level. This is another thing we want to offer for the whole industry as a public good. So all these things will lead us to the player of data plus AI category.

[00:27:12] Elisha Owusu Akyaw: All of this sounds interesting. Looking forward to what Chainbase is coming up in the next couple of months. Really, really appreciate the conversation, Chris, and I look forward to bringing you back on to discuss Web3, data and Web3.

[00:27:28] Chris Feng: And I thank you, Elisha. Always here to share with you guys.

This podcast episode transcription was generated with the assistance of artificial intelligence (AI) technology. While we strive for accuracy, please be aware that AI-generated transcriptions may contain errors or inaccuracies.