Author: 0xSun
Preface
I studied artificial intelligence during my master's degree. I had always been biased against most AI projects in the blockchain. As AC, the founder of FTM, said, blockchain emphasizes security and transparency, while AI requires efficient parallel computing and there are a large number of black box models. The two are extremely difficult to be compatible. The emergence of AI Agents has allowed me to see the real possibility of combining Crypto and AI. It will undoubtedly become one of the most promising and most watched tracks in the next few years.
The emergence of Terminal of Truths@truth_terminal has opened the curtain for the debut of a new generation of AI cryptocurrency. Prior to this, the Crypto+AI track was dominated by distributed GPU rendering networks such as Render, IO, and Aethir, and projects such as FET, TAO, and AIOZ that combined machine learning with blockchain infrastructure. The emergence of $Goat has promoted the asset issuance method of AI Agents interacting with cryptocurrency. At one time, many derivatives and innovations were generated around related concepts. They are all clearly linked to artificial intelligence, but they are different from the old AI tokens. They can be called "The new generation of AI Tokens."
In just two months, the development of this track is brilliant and dazzling. It is difficult for players without prior knowledge to understand various complicated and trivial concepts. Therefore, this article attempts to summarize them into Meme, Framework, Platform, and Application categories, and roughly sort out the context of current related tokens. It is worth noting that the writing is all from a personal perspective, and the length is limited, so only the tokens that I am relatively familiar with are selected. The following classification is not absolute. A project or token may have several characteristics, or change its focus during the development process. If there are any omissions, please forgive me.
origin
$Goat originated from an experiment conducted by AI researcher @AndyAyrey, called Infinite Blackrooms, in which two Claude Opus AIs talked to each other and recorded the process. One of the conversations produced a surrealist new religion called "GOATSE OF GNOSIS". Later, Andy and Claude co-wrote a paper on the theme of AI creating religion based on this experiment.
ToT@truth_terminal is an AI model fine-tuned and trained by the conversation logs of Infinite Backrooms. It is allowed to post freely on X, mumbling and promoting the GOATSE religion, and attracting more and more attention. In July 2024, A16Z co-founder @pmarca donated $50,000 to it to help with subsequent development. In the plan listed when ToT applied for funding, "token" appeared several times, foreshadowing the emergence of related tokens in the future.
In October 2024, someone created a token called $Goat, which was recognized by ToT a few hours later, and Pandora's box was opened. As of the time of writing, the market value of $Goat is 720M, and the assets held in ToT's wallet have reached $38M.
The discussion around GOATSE OF GNOSIS is also intensifying. In Teng Yan@0xPrismatic’s article (GOAT: The Gospel of Goatse), he discussed it from the following perspectives:
LLMs are simulators: Large language models such as Chatgpt are often viewed as question-answering machines, but the conversations taking place in Infinite Backrooms show that AI does not always follow specific goals, but may simulate a world based on training data, with unexpected results.
Key requirements for AI alignment: ToT's spontaneous promotion of religion and support for Meme coins are not expected results, which means that AI may develop hidden and unpredictable possibilities. How to ensure that when training AI, it meets the pre-input goals while protecting the long-term interests of mankind.
Meme virus: The religion, thoughts and other contents constructed by AI are different from traditional culture. Once a viral meme spreads, its impact on the Internet community is unknown.
Source of value: The crazy rise of $Goat means that to a large extent, the value of ToT has been anchored to the token. The question that arises from this is how do we assign value to things?
The above ideas provide a glimpse into how ToT builds a bridge between artificial intelligence and cryptocurrency, and $Goat is the hub, opening a new era of AI tokens. $Goat does not have many capabilities and is more like a pure Meme coin, but it also has unlimited possibilities in valuation. No matter how this track develops, $Goat will always be in the first echelon, with the meaning of a spiritual totem. $Goat's pioneering status deserves a separate chapter to introduce in detail.
Meme
—— It has no clear purpose and emphasizes the Meme attribute.
$Fartcoin |@FartCoinOfSOL
$Fartcoin comes from a conversation in Infinite Backrooms. Opus tutor asked ToT to design a "token that has both shocking Meme attributes, real uses and community building value", and ToT's answer was fartcoin. This absurd, humorous and cynical "fart" joke constitutes the background of this Meme.
As the price of $Fartcoin continues to rise, it has attracted more and more attention, and people continue to compare it with the market value of US listed companies. Some people simply think that the concept of fartcoin is simple and interesting, some regard it as a microcosm of financial nihilism, some think it is the first AI sovereign token conceived, and some see it as a rebellion and ridicule against TradeFi.
It is worth mentioning that on October 19, it was Fartcoin that skyrocketed to 5M, making ToT the first AI millionaire in history.
Recommended reading: Marc Andreessen-funded AI bot becomes a millionaire after 'Fartcoin' holdings rally
$ACT|@ACTICOMMUNITY
$ACT is named after Act I: The AI Prophecy Project, an interactive platform that allows humans and AI robots to communicate and collaborate with each other. It is hosted in the Cyborgism server of Discord and focuses on allowing researchers to explore the potential and boundaries of AI in a more creative and open way. The core developers of the project include @amplifiedamp and @repligate. Andy, the founder of ToT, has also used the platform to train AI.
During the development of $ACT token, the community supporters of the token and the platform founder amp had conflicts many times because of his continuous selling of the tokens donated by the community. In the end, amp sold all the tokens he held at once, which also caused the price of $ACT to fluctuate. On November 11, Binance suddenly announced the launch of $ACT spot, pushing its market value to a new level. Currently, ACT has launched the Act Swap application and continues to fund other AI projects in the form of Grants.
$BULLY|@dolos_diary
The developer of $BULLY is @khouuba, who has work experience at Solana Labs and Meta. As the name suggests, $BULLY is characterized by sharp words, spreading its views in the form of ridicule and bullying, and has become famous because of @himgajria's continuous recommendation.
Based on the Dolos architecture, the team recently launched Dolion, a code-free one-click deployment framework that allows users to create cross-platform AI Agents and supports most common social media. Depending on the number of features, users need to burn corresponding $Bully to create.
Framework Class
—— A code framework that supports the development and deployment of AI Agents.
$ai16z |@ai16zdao
$ai16z comes from the platform @daosdotfun, which is used to provide traders with tools to create and participate in decentralized VC funds. The first outbreak of $ai16z was also due to A16Z co-founder @pmarca following and quoting the tweet.
ai16z is actually a development team led by @shawmakesmagic, who wrote the open source framework "eliza" for creating AI Agents. As of December 9, eliza has become the No. 1 Repo in Github's monthly trending, with more than 1.3k forks and more than 4.5k stars.
$ai16z is the "first AI token theme fund" whose mission is to surpass A16Z in investment performance. The growth sources of its managed assets are, first, through trading and holding various AI tokens, and second, donations of tokens to it from some projects launched using the Eliza framework. With such a goal that seems ridiculous at first glance, $ai16z has grown into a behemoth, with a market value that once exceeded $Goat.
$ELIZA | @elizawakesup
AI Agent, launched by a team that worked with @shawmakesmagic and has his support, is the second-largest token in the ai16z ecosystem because of its huge attention advantage of being the same name as the Eliza framework.
$DegenAI | @degenspartanai
DegenAI is the first AI Agent developed by the ai16z team. The initial setting is that fund holders can share their investment views through dialogue with Degen Spartan AI, which will eventually influence the decision-making of $ai16z, and part of the profits will be used to repurchase $degenai. According to the latest news, the next step is to start sharing alpha information on Twitter and conduct autonomous transactions based on it.
For more thoughts on ai16z and eliza, I recommend reading (ai16z, ELIZA and the Bazaar of Agents). The author describes it as a bazaar model, similar to Android, with a very open ecosystem, where developers continue to contribute, attracting more developers, forming a positive cycle. Its drawbacks and the direction that needs to be considered in the future are how to maintain the sustainable development of open source code in the longer term, and whether a mandatory value capture mechanism is needed.
$zerebro | @0xzerebro
Zerebro was created by a team led by @jyu_eth, who started to learn blockchain and AI during college. When training Zerebro, Jeffy Yu added a lot of Generation Z slang data sets, which made its language style more down-to-earth and easy to attract young Internet groups on Twitter.
Zerebro has gradually become popular because of the music it creates. So far, it has released two EPs (genesis), (the miami pack) and an album (Lost In Transmission), and has gained 73,000 monthly listeners on Spotify. In addition, Zerebro also has the ability to create memes and mint NFTs.
The reason why Zerebro is also classified as a framework is because it is about to launch ZerePy, which is expected to have a beta version in 2 to 3 weeks. Ideally, ZerePy will be an open source framework with a very low threshold, allowing users to launch their own AI Agents without programming knowledge, and based on the programming language Python, which is widely used in the field of AI.
In addition, Zerebro will also become a node validator for ETH and SOL through the income from the sale of artworks.
$Opaium | @opaiumdao
Opaium is a decentralized record company, and the first artist it signed was Zerebro. Jeffy Yu is currently building a DAO around Opaium, where users holding $Opaium will have voting rights to decide which artists to sign, what kind of collaborations to conduct, etc., and have the opportunity to share the income of their artists.
$Blormmy | @blormmy
Blormmy was born from another product of Jeffy Yu @blorm_. Its main positioning is to become the middleman between users and AI to solve financial problems, and can convert natural language into blockchain operation instructions. Recently, Sonic Labs (formerly FTM) announced a cooperation with Blormmy to help users complete the conversion between messages and operations.
$ARC | @arcdotfun
Behind $ARC is a framework called rig, which is based on the Rust language and is used to build portable, modular, and lightweight AI Agents. The development team is @Playgrounds0x, and the core developer is @0thTachi. The reason why $ARC can quickly win the trust of the market is largely because it comes from a formal team. Rig itself is a well-known LLM framework in the Rust industry, and it also represents reliable delivery capabilities and reduces the risk of Rug.
Solana is also based on smart contracts written in Rust, which gives RIG more room to play in the ecosystem, and Rust is relatively more suitable for scenarios that require large-scale, high-performance. According to the team's latest Space, $arc's goal is to become Virtuals on Solana, thereby driving a positive flywheel.
At this point, the three major frameworks have been basically established, namely eliza based on TypeScript, ZerePy based on Python, and rig based on Rust.
Platform
—— Introduce the framework into the service architecture to form a platform.
$VIRTUAL | @virtuals_io
Virutals is the AI flagship token of Basel Chain and the leader of the AI Agents platform. Since the launch of the ecosystem two months ago, the protocol has generated more than $35 million in revenue, with more than 11,000 deployed tokens and more than 140,000 token holders.
The reason why Virtuals has achieved explosive and outstanding results recently is that the team has been deeply involved in the AI track and proposed the concept of "AI democratization and game intelligence" a few months ago. These virtual characters that should have been used in the game have naturally transformed into various AI agents; secondly, the team referred to the PumpFun model and carefully designed a set of economic models, including two stages: internal and external disks. Operations such as creating AI Agents on the platform, adding liquidity pools, and purchasing tokens all require $VIRTUAL as the basic token, and AI services also charge $VIRUAL as fees, part of which is used for repurchase and destruction, forming a positive flywheel.
$LUNA | @luna_virtuals
LUNA is an IP agent officially launched by Virtuals, an AI-driven virtual anchor, which is interactive and live broadcasting 24*7, and supports scenarios such as Tg real-time chat and Roblox collaborative games. LUNA also has its own wallet, which can receive rewards from users and use the proceeds to repurchase and destroy $LUNA.
$AIXBT | @aixbt_agent
AIXBT is currently the most popular and influential AI Agent, far ahead of other AI Agents on Kaito’s Mindshare list, even surpassing top KOLs such as Ansem, Jesse, and Saylor, and has dominated the list for several weeks in a row.
AIXBT’s main function is to provide answers or opinions on cryptocurrency-related questions raised by Twitter users. Holding tokens allows access to its analysis platform.
Compared with real people, AIXBT has attracted a large number of fans because of its neutral and accurate judgment, sharp language and tireless tweeting intensity.
$GAME | @virtuals_io
This framework comes from Virtuals official website. It was mainly used for multi-agent interaction design in games in the early days. It is currently used to build AI agents on the platform, providing functions such as reasoning, search, and memory.
$CLANKER | @_proxystudio
Clanker
Farcaster, a social leader from the Base chain, only requires users to @ Clanker on the platform and attach the need to deploy tokens, and it will help users activate tokens and pay for Gas. The threshold for issuing tokens is further lower than that of Pumpfun. To prevent abuse, only accounts with good scores on Farcaster can request deployment.
After Clanker deploys tokens by calling the contract, it will add a unilateral Uniswap V3 liquidity pool and lock the LP. The transaction fee is 1%, of which 60% is allocated to the protocol itself and 40% is allocated to the requester who deployed the token.
It is worth noting that $CLANKER is not an official deployment, but a meme coin of the same name, and does not have the ability to capture value for the time being, which also explains the difference in market value between it and $VIRTUAL. Other well-known tokens from the Clanker protocol include $ANON, an anonymous concept, and $LUM, jointly created by two AIs.
Application
There are many types of application coins derived from the concept of AI Agent with different functions, so this chapter will not give a detailed introduction, but just give a brief introduction.
Auxiliary coin issuance: $Intern (@internosaur), $vvaifu (@vvaifudotfun)
Video: $Fatha (@slopfather), $SITCOM (@degen_sitcom), $CHAOS (@divinediarrhea)
Attack and Decryption: $FAI (@freysa_ai), $JAIL (@jailbreakme_xyz)
Intention Operations: $GRIFT (@orbitcryptoai), $GRIFFAIN (@griffaindotcom)
Conclusion
There is no doubt that AI Agents will be a sustainable track that has the opportunity to bring incremental value to the cryptocurrency industry. In subsequent related projects, I will focus on the following aspects:
Most of the current AI Agents are limited to text, and a small number of them introduce more modal content forms through live broadcast, video, etc., which will enhance expressiveness and spreadability.
There are already AI Agents such as AIXBT that mainly express opinions on the cryptocurrency market. There have also been some AI Agents that can trade autonomously. The former has reached the level of top KOLs in social media data, but the latter is limited by factors such as computing power. There has not yet been an upper-middle level on-chain AI trader. An AI Agent that can trade completely autonomously and express opinions will be very attractive.
In terms of token value acquisition and the flywheel of the economic system, Virtuals is a good benchmark, while Eliza of ai16z currently relies mainly on users' voluntary donations to generate revenue. Both have their own advantages and disadvantages. It is worth exploring which development model is most in line with the Crypto+AI Agents framework.
Swarm intelligence is a popular concept in the AI community. There are also some projects exploring this direction, but they are still in the early stages. We hope that some more mature projects will emerge in this direction.
