Dialogue with GaiaNet CEO: Challenging Giants, Creating a Decentralized "ChatGPT"

Interviewer: flowie, kit, ChainCatcher

Guest: Matt Wright, Co-founder and CEO of GaiaNet

Editor: Marco, ChainCatcher

In 2015, Matt Wright became involved in crypto after helping the UK's leading bank, Barclays, host a blockchain hackathon, and he has been in the crypto space for nearly a decade now.

Matt Wright previously worked at JPMorgan, where he was involved in developing the open-source blockchain platform Quorum. After Quorum was acquired by Consensys, Matt joined Consensys as the community lead and later managed Consensys's DAO organization and its accelerator, Fellowship.

Most of the time, Matt Wright has been interacting with developers. With the advancement of AI technology, more and more developers around him are getting involved in AI-related projects.

However, Matt soon realized that the current AI field is plagued by many issues due to being dominated by a few centralized organizations in Silicon Valley, such as the censorship and bias of closed-source models, high model training costs, user privacy, and IP ownership, among others.

Therefore, Matt decided to lead his team to create a truly democratic AI infrastructure that could solve these problems, and Gaia was born in May of this year. As an open-source distributed AI infrastructure project, Gaia aims to be a decentralized alternative to AI giants, intending to decentralize AI Agent software so that AI data content providers can have rightful ownership and compensation for their data content.

The first use case of GaiaNet emerged from academia. Dr. Yang, the director of the FHL Vive Center at the University of California, Berkeley, is using GaiaNet's distributed network to create a decentralized teaching assistant by leveraging a wealth of research and data, including his courses, writings, and student feedback.

In addition to building network infrastructure, GaiaNet is also seeking collaboration with academic institutions that possess valuable data, aiming to help them create their own AI Agents.

At its inception, GaiaNet secured $10 million in seed funding, with key strategic advisors including Lex Sokolin from Generative Ventures, Brian Johnson from Republic Capital, Shawn Ng from 7RIDGE, Kishore Bhatia, EVM Capital, Mantle EcoFund, and ByteTrade Lab.

Veteran "Player" of Hackathons

1. ChainCatcher: How did you first get into the crypto space? What other experiences in the crypto field did you have before founding GaiaNet?

Matt Wright: My career started with an international hackathon company and accelerator called AngelHack. AngelHack organizes over 200 hackathons each year in more than 60 countries or regions, including North America, Latin America, and mainland China.

I was involved in AngelHack's initiatives focusing on emerging technologies and markets, organizing hackathons for over 20 Fortune 500 companies, such as Mastercard, AWS, ABInBev, Capital One, Smartone, Motorola, and Barclays.

During this time, I also contributed to AngelHack's strategy in mainland China, hosting several events in Shanghai and Shenzhen. I really enjoy and understand the culture of the Chinese-speaking regions.

Around 2015, I participated in a blockchain hackathon hosted by Barclays in New York.

In our discussions, they told me that blockchain was disrupting banking. Although I had heard of Bitcoin before, this was my first encounter with engineers building blockchain applications and infrastructure.

The blockchain use cases they mentioned were topics I cared deeply about, such as the transparency and democratization of open-source technology, as well as restoring people's sovereignty and assets that had been stripped away. This experience prompted me to learn more and eventually join the crypto field.

Around 2017, I joined JPMorgan, participating in the development of their open-source blockchain platform Quorum, which is a fork of the Go Ethereum client (geth) designed to handle private transactions with a group of known, permissioned participants.

Soon after, Quorum was acquired by Consensys, and I joined Consensys. They initially wanted me to take a marketing role, but I was determined to focus on the developer community, hoping to apply my organizational experience from AngelHack at Consensys.

Later, I became the head of DAOs at Consensys, thinking about how to build applications for on-chain organizations that have no CEO and how to use the community for tokenization incentives, among other things.

I also managed Consensys's accelerator, Fellowship, providing support for some outstanding builders.

2. ChainCatcher: You also founded an investment firm, EVM Capital. Why did you decide to venture into institutional investing?

Matt Wright: We interacted with a large number of developers, many of whom spent a lot of time explaining their products to VCs who might not fully understand the technology, and they often lacked patience.

Since we have been dealing with developers, we saw this opportunity and wanted to help many promising products capitalize and find their value networks.

Currently, we view EVM Capital almost as a fund for GaiaNet. If our portfolio focuses on AI, governance automation, identity, and reputation automation, then we can recommend GaiaNet's products to them for synergy.

Challenging Giants, Creating a Decentralized "ChatGPT"

3. ChainCatcher: You co-founded GaiaNet around May this year. How did this entrepreneurial idea come about? What backgrounds do the core team members have?

Matt Wright: The core team of GaiaNet mainly comes from leading projects like Consensys, EVM Capital, and Aleo. One of GaiaNet's co-founders, Sydney Lai, was the head of developer relations at companies like Magic, Mantle, Filecoin, and Outsystems, and she is very influential in the developer community. She is now our head of development.

Another co-founder, Shishank Shipada, has extensive experience in institutional capital and multiple entrepreneurial ventures, as if he has lived two lifetimes. He has co-founded and exited three companies and has experience managing investments for several institutional capital firms. He currently runs a family office with a scale of $7 billion and has co-founded a $600 million fund.

I chose to establish Gaia to promote the democratization of AI technology. Web3 allows anyone to become their own bank or knowledge monetization service provider, and AI is enabling anyone to become their own content generator or intelligent agent. In the future Web3 world, the number of participating AI bots may exceed that of real humans.

However, currently, AI data is dominated by a few centralized organizations in Silicon Valley. If we continue to hand over control and trust to a centralized service provider, the consequences are unimaginable.

4. ChainCatcher: What specific solutions does GaiaNet offer to address the centralization issues in AI? You mentioned in your press release that you will first target the needs of higher education. Can you provide examples of how GaiaNet addresses problems in this field?

Matt Wright: Our first cool use case started with Dr. Yang, the director of the FHL Vive Center at the University of California, Berkeley. He has a decentralized teaching assistant plan.

Currently, in the higher education sector, especially in addressing high-enrollment courses (like introductory computer science and engineering), there is a significant shortage of teaching assistants.

Dr. Yang is placing a wealth of research and data, including his courses, writings, project white papers, and student feedback, into his GaiaNet node, with each GaiaNet node acting as an AI Agent similar to a real-world teaching assistant.

The approximately 30 PhD students he supervises will communicate with the instructors at the FHL Vive Center through this intelligent teaching assistant.

With long-term accumulation, it is very imaginative that if someone wants to co-author a paper with this intelligent teaching assistant, the assistant could receive compensation. This means there could be programmable royalties.

Currently, we support various open-source large language models and have established corresponding core nodes in the Gaia network, such as Meta's llama-3-8b, Google's Gemma and CodeGemma, Microsoft's Phi-3-mini model, Alibaba's Qwen series models, and more.

We allow users to use their proprietary data for training, vectorizing the data for vector storage.

When an application runs, our nodes can operate in different environments with the same performance. Users only need to enter a few lines in the command line to run their own Agent node. They can choose to host it on a GPU, local computer, or in the cloud.

5. ChainCatcher: Besides the higher education sector, what other application scenarios will GaiaNet expand into in the future?

Matt Wright: We are looking for scholars and institutions or individuals with a wealth of proprietary data, and we hope to help them create decentralized AI Agents. I believe this is our greatest advantage.

For example, companies like The New York Times are suing OpenAI because OpenAI is training models based on The New York Times' hundreds of years of newspaper data. In this case, we can actually help The New York Times have its own AI.

At the same time, through Web3 technology, this AI is open-source and permissionless, allowing users who need it to call upon it at any time and compensating The New York Times through token economics.

6. ChainCatcher: GaiaNet secured $10 million in seed funding at its inception. What do you think is the core reason for GaiaNet receiving their financial support?

Matt Wright: On one hand, the direction of AI + Web3 that we are pursuing is something they are optimistic about in the long term.

On the other hand, our core team's comprehensive capabilities are well-regarded. Most of the core members of the EVM Capital and GaiaNet teams come from Consensys, Filecoin, and Aleo, bringing together technical expertise, developer resources, and rich experience in commercialization and capital operations.

Over 17,000 Registered Nodes

7. ChainCatcher: Recently, GaiaNet launched its Beta product. What updates does it have compared to the Alpha testing phase? How is the operational data?

Matt Wright: A significant improvement in the Beta version is that it allows nodes to register to wallet addresses, so each time a node starts, it operates like opening and controlling a network node through a wallet. We currently have over 17,000 registered nodes and are continuously expanding.

8. ChainCatcher: What is the roadmap for GaiaNet moving forward?

Matt Wright: On the technical front, the primary goal is to improve performance to ensure that nodes are very useful to our community, whether these nodes are in the cloud or on GPUs.

We are also in contact with some teams building GPUs to understand how they work in different, smaller capacities.

Additionally, we are researching domain names, node staking infrastructure, etc. Some of our models require training on large amounts of data, and we are exploring how to ensure the accuracy of verification data and the security and effectiveness of the Gaia network without accessing the data through technologies like zk proofs, DA, and Eigen AVS.

In summary, we are trying to maintain our open-source Web3 infrastructure with different cryptoeconomics, ensuring it operates with privacy protection and operational efficiency like any centralized company while guaranteeing that participants can receive compensation for their applications.

Besides technology, we need to attract more developers into our ecosystem. We have already discussed integration with over 200 AI software companies, and I believe that in the future, we will have a plugin ecosystem that can bring many different functionalities to the nodes within the ecosystem.

9. ChainCatcher: What is the biggest challenge GaiaNet is currently facing?

Matt Wright: I believe the biggest challenge right now is how to maintain good collaboration among teams spread across different time zones. Some of our teams are located in New York and Los Angeles, while others are in Singapore and Taiwan.

But I believe we can handle it well. I have spent a lot of time in the Asia-Pacific region, including Hong Kong, and have collaborated with teams from different cultural backgrounds. We are now overcoming the challenge of merging the two cultures that are most deeply involved in Web3.

10. ChainCatcher: Regarding AI + Web3, what other interesting innovative use cases do you see?

Matt Wright: We are currently researching a very interesting use case.

We are collaborating with a decentralized governance team called Boardroom, which has a wealth of on-chain governance data.

We are thinking about how to effectively build Gaia's DAO and Foundation to create snapshot proposals and statistical proposals.

It's important to note that this ecosystem manages approximately $40 billion in capital, and participants make decisions around this capital, with thousands of proposals being issued every month.

We hope to help build a GPT supported by this data, allowing users to interact directly with the intelligent agent and make quick proposal decisions.