Exclusive interview with Artela CEO Jerry: Building the first full-chain AI high-performance public chain

Interviewer: Mensh, ChainCatcher

Guest: Jerry, CEO of Artela Network

Artela Network is the first high-performance public chain that supports full-chain AI. On-chain scalability, flexible block space, parallel execution and modular architecture make Artela the "natural choice" infrastructure for developing decentralized AI agents. Artela makes AI behavior highly verifiable and scalable, making a complex, independent, and transparent AI possible.

The community uses "an unkillable monkey" to refer to such an AI agent that can live forever across devices. On January 16, 2025, the Artela mainnet will be launched. Every developer will have the opportunity to create an "unkillable monkey". monkey".

ChainCatcher is fortunate to have Jerry, CEO of Artela Network. Jerry has rich engineering and development experience. He was the former R&D director of Apple China and the former chief architect of Ant Chain. With the pursuit of innovation as the main line of his career, he decided to leave Ant in 2022, bringing with him the technical experience and deep cultivation in the industry. Insight and build a new high-performance public chain.

In this interview, we talked with Jerry about the technical details of Artela, the development process of star AI agents on the Internet, and the development prospects and hidden dangers of AI agents.

**Starting with performance, the star team builds a full-chain AI

infrastructure**

ChainCatcher: When Artela is first established in late 2022, what product do you want to build? As the first public chain to support full-chain AI agents, what changes have occurred?

Jerry: When we were founded, we wanted to build a high-performance public chain. Artela’s blockchain technology stack was designed from the ground up. But we are not creating wheels out of thin air. On the one hand, our team has accumulated a lot of technology in Ant Chain before. Our core technical developers and I are the core architects of Ant Chain. At the same time, we don’t want to reinvent the wheel. We also want to build a public chain with better connectivity with existing facilities, so we chose EVM compatibility as the baseline, and also chose to use the Cosmos SDK. Between these two On this basis, we quickly put together the basic chain, and then added our previous experiences and lessons to re-develop such a set of public chains.

Later we discovered that we still had to make an application to reach users. There have been too many reinventions in the blockchain industry, and too much money has been invested in building infrastructure, but nothing truly practical has been brought to users. Then we have been looking for a suitable application until the last three or four months when we found AI agent. We have fully invested in this and combined it with AI on the basis of high-performance public chains. I think we have now found a better breakthrough point. ,

ChainCatcher: Compared with other AI agent development platforms such as Virtuals, what are the advantages of Artela?

Jerry: Artela has some unique technical advantages that enable it to support truly decentralized and fully autonomous AI agents. First, because Artela 's technology stack is designed from scratch, it can achieve high performance and low cost , allowing the AI ​​Agent to run efficiently without incurring excessive costs. The second is Artela 's original aspect programming and flexible block space technology, which allows AI agents to freely expand and operate on the chain. Third, Artela provides a decentralized Trusted Execution Environment (TEE) to ensure the privacy of AI agents and the reliability of execution, which is very different from the centralized execution methods of many existing platforms.

ChainCatcher: Let’s look at them one by one. In what specific aspects does Artela's technology stack achieve high performance and low cost?

Jerry: First, for the chain itself, we have made many detailed optimizations through full-stack parallelization, including storage calculations and network consensus, to reach tens of thousands of TPS (transaction processing per second). In a production environment, this is the direction that new high-performance public chains should go. Solana or Layer 2 of Ethereum, its TPS in the production environment is 500-600, and we need to increase it to tens of thousands.

In addition, we reduce costs in the process. Some public chains are very centralized in order to improve performance and have very high hardware requirements for nodes. Artela has moderate hardware requirements for nodes while achieving very good performance.

In the past two or three months, we have done a lot of work on the integration of AI agents. We found that having a chain alone is not enough, and we also need to find application categories that integrate with the market. AI agent is particularly suitable for Artela's technology and is optimistic about its development prospects. We feel that the AI ​​agent is not a plate rotation, but a long slope with thick snow for a long time, so we integrated the native large language model (LLM). This is also because Artela's chain supports multiple VMs. More complex calculations can be moved to the chain. We also defined an AI agent account, so that the core functions of the AI ​​agent can be put on the chain. This is a big breakthrough for the existing market, reaching chain-native AI.

ChainCatcher: Artela’s two biggest technological innovations, Aspect programming and flexible block space, can you introduce them in detail?

Jerry: Aspect is a development environment that supports more languages. It can attract many Web2 programmers because it supports more languages ​​such as Assembly, Rust, and C++. It is a more powerful and flexible system . When working on Ant, everyone felt that it was a better programming system that could be used for large-scale applications.

Flexible block space is a practical performance for large-scale applications. For example, Alipay will have an instantaneous transaction volume peak at zero o'clock on Double Eleven. So the design must have this kind of flexibility. Normally, the trading volume is not that large, but during a specific period, the trading volume may increase rapidly. Flexible block space solves this problem, and it can be expanded more flexibly. Applications like AI agents or games have timeliness requirements. It can meet this need through flexible expansion.

ChainCatcher: What differences does TEE bring to AI agents in terms of privacy and execution reliability?

Jerry: TEE is a hardware environment that can be verified from the outside, but the encryption execution inside is a black box. When there is an AI agent to manage the user's private key, the secret information will not be leak, but at the same time you can verify permissions, so we will deploy TEE nodes on the chain, and then implement such a hardware encryption environment, which is very efficient.

But beyond that, we remain open to other encryption technologies. In addition to TEE, there are also different encryption methods such as CTML and FHE. In the future, we think there will be opportunities to integrate them into the chain and do encryption at different levels and in different environments.

ChainCatcher: Since it is hardware verification, will it be decentralized enough?

Jerry: Not all network nodes have TEE, so on the one hand we will deploy a small number of TEE nodes, and on the other hand we will also find TEE resources with partners to decentralize it.

In addition, we recently conducted a very interesting social experiment called an unkillable monkey. When an AI agent is on the chain, it can gain immortality. Even after the TEE environment is damaged, the corresponding data can still be recovered from the chain. The concept of unkillable monkeys has been very popular in the market, and the community has even issued a community coin for foc AI.

**From community assistant to wallet manager, the present and future of

AI Agent**

ChainCatcher: What is the relationship between Artela and Eliza?

Jerry: Eliza framework is actually an open source AI agent framework developed by AI16Z. I am a partner involved in development and community construction. If you look at its underlying code, there are very few parts about blockchain, so we See a space that combines AI with blockchain. What we do with foc AI is to build a series of plug-ins that are integrated with the blockchain on the original basis. So in a broad sense, what we are doing is also part of the Eliza framework, focusing more on the part that is integrated with the blockchain.

Eliza provides us with a powerful infrastructure that provides a trustless execution environment. By combining Trusted Execution Environment (TEE) and Zero-Knowledge Machine Learning (ZKML), every behavior of the AI ​​agent is trustworthy and cannot be tampered with. In addition, Eliza also enables AI agents to complete complex on-chain tasks without manual intervention, which is consistent with our expectations for AI agent independence.

Since Eliza is open source, we have received support from the Eliza community. The organizers of the Eliza community came to Asia to communicate with our team, and some developers from the Eliza community are now involved in our work.

ChainCatcher: Which star AI agents are running on Artela now?

Jerry: The most popular is AI Vinci. It has over 50,000 followers on Twitter. We started building a testnet in the summer, using it as a chatbot to communicate with users, explain Artela knowledge points, or assign tasks to everyone, so it was still relatively rough at that time.

After the Eliza framework came out, we quickly implemented it on the Eliza framework, and its intelligence has been greatly improved. Our mainnet will be online soon, and we will move AI Vinci to Artela's mainnet and use all the basic functions of Foc we made. There are two core features. One is that it can achieve background verification. Using the background link can verify that every user interaction is automatically generated by AI. The other is to realize the immortality function on the chain. Through the combination with TEE and Artela main network, eternity can be truly realized on the chain.

We are conducting a social experiment today - electing Artela's community manager. We now have two candidates, one is AI Vinci, and the other is a young partner who has been active in the community for more than a year. Preaching to Artela very carefully. We want to let the community vote to choose a community manager between humans and AI. Results will be released around January 16th.

ChainCatcher: What is the training method for AI Vinci?

Jerry: We use Eliza's framework to provide it with a large amount of data in related fields, including Artela's knowledge base and full-chain AI-related knowledge base. In addition, it continuously interacts with the community. In the process, it Slowly, a certain personality of my own has emerged, including a sense of humor.

Currently it is defined as a community manager, but it may be able to do more work in the future, such as not only serving our community, we can clone an AI to run other communities. If it continues to develop, it can be given more on-chain capabilities. For example, it can reward you, provide you with economic interactions, and form a set of economic models.

ChainCatcher: AI agents now have many application scenarios and cases, such as tweeting, online idols, or managing communities, etc. Among so many application cases, which application directions are you most optimistic about?

Jerry: Today’s AI agents are still in the “Stone Age”, and homogeneity is very serious. The next step I am more optimistic about is the combination with the chain and the development of full-chain AI. A large application category is AIFi, which uses AI to manage wallets and then operate various assets on the chain. For example, if Ethereum is below $3,000, it will buy it. But it can also be used for more complex operations, such as YouTube farming MEV, and arbitrage between various contracts. When the ability of the AI ​​agent to combine with the chain is strong enough, it can be used in a financial market that operates 24 hours a day. run.

ChainCatcher: Speaking of AI agents participating in community governance, imagine a future where AI is powerful enough. It is difficult for us to tell whether the AI ​​agent is malicious. Its independent governance will bring some risks to the governance of the community. Regarding the increasing authority of AI agents, What do you think of the security issues that will arise later?

Jerry: This is a big question. What we imagine is that in the future, 99% of the actions on the chain will be operated by AI Agents. To some extent, the boundaries between humans and AI have become blurred. It is becoming more and more powerful and intelligent, even surpassing far more than humans. In many professional fields, you don’t know that the situation you just mentioned will arise, just like what if there is a bad person in a group? I don't have a too complete answer, but I'll keep it verifiable and transparent anyway, as well as some of its fundamental capabilities.

**Achieve more complex on-chain interactions, leading to a broad market

of 10x+**

ChainCatcher: What does Artela’s roadmap look like in the new year?

Jerry: In the new year, we will focus on three areas. The first thing is to improve the performance of all aspects of the blockchain, decentralize it, and secure it, so that it can be verified on the public network. The second thing is to work with the open source community to build the AI ​​architecture. The third thing is to work with developers to create applications for retail investors. We have corresponding incentive plans to support developers.

ChainCatcher: Artela is about to launch its mainnet and needs to attract more developers. So what activities are you currently doing to attract more developers to the Artela ecosystem?

Jerry: After the mainnet is launched, we will launch a platform for ordinary developers. We need to build end-to-end products based on the platform, so that retail investors can also participate. So next we will definitely work hard on the product side. Artela previously held an event called Renaissance, which we did last summer. The origin of this name is that we believe that AI is developing faster and faster, posing great challenges to the development of the entire human society. The last such big technological progress was during the Renaissance, and the central idea of ​​the Renaissance was people-oriented.

When AI becomes powerful, people will lose their jobs, people will lose their beliefs, and people's lives will become meaningless. So how can we put people at the center again without being influenced by technology? This is our original intention in the Renaissance Project. We specifically create content to encourage more users to interact on the Artela testnet.

ChainCatcher: What other performance needs to be optimized in TEE Eliza currently?

Jerry: Technically speaking, it is to enhance verifiability. Currently, all operations performed by Eliza within the TEE require an explicit logging mechanism to ensure that these logs are available for auditing and verification at every step. The second is to enhance the ability to interact with the blockchain. Currently, Eliza needs to rely on some blockchain components during execution, such as trusted blockchain clients and data synchronizers. These need to be further developed within TEE to ensure that the AI ​​Agent can interact with the blockchain more smoothly. interact with the chain system.

But I think the real test is how users can trust AI to help them manage their money. The first is whether it can be managed, and the second is whether it can be managed well. What we have to do now is the first step, so that he has the ability to manage it. I think it will be done soon. Can you manage it well next? Will it lie to you? How to deal with attacks against AI Agents? Can it give you the best trading strategy? In extreme market conditions, can it make corresponding adjustments? I think there are still many more advanced challenges.

ChainCatcher: In 2021, the domestic Internet has also set off a wave of AI virtual human project boom, but so far it seems that it has not been accepted by the market. What is the difference between the AI ​​agent craze in Web 3 and the AI ​​virtual human in Web 2?

Jerry: I think technology develops in cycles and will explode at the right time. There have been three technological winters in AI in history. Autonomous driving has actually been implemented in the United States. Web 2 virtual humans were proposed in the last cycle and had a short-lived explosion. Later, it was discovered that the technology was not yet mature, but it does not mean that it has no value. It will really explode when the time is right. Web 3 is an industry that is developing and iterating very quickly. It is not the same as Web 2. There are many small teams constantly trying, so many new breakthroughs are likely to occur in Web 3. Including this outbreak of AI agents, it has also attracted the attention of many traditional Web 2 AI companies. They have also found that combining with blockchain can provide a catalyst for the development of AI. It can help AI provide funds to quickly reach users in the form of tokens.

ChainCatcher: Which one has more advantages between the open source AI agent framework and the closed AI agent framework?

Jerry: I think the current business models have their own merits. It’s a bit like the analogy between iOS and Android. Both iOS and Android are quite successful. iOS is a typical closed-source ecosystem, and relatively speaking, the value obtained is higher. Android has a wider range of users and is developing well. In Web 3, everyone may be more inclined to open source. Therefore, our strategic choice is to embrace the open source community. I think we will see more projects that were not open source before taking the open source path.

It's more of a cultural consideration than a technical one. Because open source can unite more developers and allow more people to participate. I think Web 3 is more about how to make grassroots communities bigger, which will have more vitality.

ChainCatcher: The arrival of AI agent will be like the DeFi Summer in 2020, focusing on the token issuance, custody, and trading of AI agent; infrastructure construction, framework, standards, communication, stacking, and SaaS packaging; application development, operation, aggregation, and iteration , landing, etc. will create a huge new "bubble" narrative?

Jerry: The first judgment is that the AI ​​agent will be more than 10 times, or even 100 times larger than DeFi Summer. AI is now in the "Stone Age" and it is too rudimentary. What will lead to future prosperity is the realization of a complex transaction system on the chain. It may also exist in another form, and its interaction with people will be different, and it may not require the existence of Twitter. It is an AI agent in a broad sense. This year should be the first year of AI agent. The current overall market value is 3 billion US dollars. It is normal for it to increase by 100 times this year.