DeFAI: New Opportunities for AI Agents

At the beginning of January, the AI Agent projects on the blockchain were gaining momentum. The market capitalization of Virtuals and ai16z reached $5 billion and $2.5 billion respectively, while underlying framework projects like ARC and Swarms exceeded $600 million in market capitalization. DeFAI projects such as BUZZ and Grift touched $200 million in market cap. AI Agent projects from the Web2 domain, such as Pippin and AIOS, also entered the crypto market to issue tokens. The market is generally optimistic about the “AI Agent revolution” and looks forward to the arrival of the “DeFAI spring.”

However, as market sentiment shifted, the liquidity of AI Agent tokens rapidly dried up. The emergence of DeepSeek further burst the AI bubble. The market values of Virtuals and ai16z fell below $1 billion and $500 million respectively, while other AI Agent tokens generally experienced declines of over 90%. The AI Agent sector plunged into a harsh winter, with the grim reality of applications struggling to take off and deeply trapped investors.

Due to the relatively small overall size of the AI Agent sector (less than 7 billion USD), many people have limited understanding of it. This article will first introduce the AI Agent sector, and then analyze why DeFAI is expected to become the second growth curve for AI Agents, potentially bringing a market opportunity worth hundreds of billions.

Overview of AI Agent Section

AI Agent is an intelligent entity capable of autonomous actions, executing tasks, and interacting with external tools.

The core of the AI Agent is model services, with main participants including ChatGPT, Anthropic, Gemini, DeepSeek, and Llama, among others. However, the AI Agent is not just a conversational language model; it also needs to manage its own state (such as dialogue history and memory), invoke various tools, and execute safely.

In the Web2 domain, AI Agent projects mainly focus on enterprise services. In the Web3 domain, the main types include AI Agent launch platforms, frameworks, infrastructure, agents, and DeFAI.

Launch Platform

The launch platform is currently the type of project where AI Agents can capture value most easily. It allows users to create, deploy, and monetize AI Agents, similar to pump.fun in Meme coins, but focuses on AI Agents.

Virtuals is currently the largest launch platform, launched on the Base network. It has issued over 100,000 Agents, including well-known projects such as virtual anchor Luna, market analyst AIXBT, and the Agent development framework G.A.M.E.

The Virtuals team was established in 2021, originally as the gaming guild PathDAO. The release of GPT in 2023 shifted the team’s focus to the AI field.

framework

The framework is a toolkit for developing AI Agents. Solana’s official list includes five main framework projects: SendAI’s SolanaAgentKit, ai16z’s Eliza, ARC’s Rig, Crossmint’s GOAT, and Zerebro’s ZerePy.

ai16z is currently the second highest market capitalization AI Agent project. In addition to providing the Eliza framework, ai16z is also a venture capital fund in the AI field, and it plans to develop into a Layer 1 public blockchain for AI Agents in the future.

ai16z was founded by @shawmakesmagic on the Daos.fun platform. The core concept of the project is to use AI models to simulate the investment decisions of the well-known venture capital firm a16z and its co-founder Marc Andreessen, combined with the suggestions of DAO members for investment. This innovative model aims to combine AI trading strategies and decentralized governance to enhance the transparency and credibility of investments.

Infrastructure

The infrastructure is an off-chain platform or tool built around AI Agents, covering a wide range. Examples include the AI Agent data aggregation service cookie.fun, the testing platform Jailbreak, the resource-sharing network FXN, the model fine-tuning training service LUMO, and the AI Agent Layer 1 public chain Artela.

cookie.fun tracks over 1,000 AI Agents, building an index and data layer that supports displaying metrics such as mind share, smart follow, engagement, and on-chain data. Its core function is to aggregate data sources, allowing all AI Agents to compare and compete in various aspects.

Jailbreak aims to create a decentralized platform that allows companies to test AI models and agents in a distributed environment. By identifying vulnerabilities before deployment, it is dedicated to enhancing the security of AI models in real-world applications.

FXN is an Agent resource sharing network that uses a unified decentralized protocol to connect all frameworks’ Agents. The core of FXN is a P2P resource sharing protocol that enables Agents to securely share functionalities.

Lumo Labs specializes in developing custom AI models for the Solana ecosystem, which can assist with market analysis, risk assessment, and project planning related to Solana.

Artela is the first high-performance public chain that supports full-chain AI, featuring elastic block space, parallel execution, and a modular architecture, which provides strong verifiability and scalability for AI behavior, making it possible for complex, independent, and transparent AI.

Intelligent Body

The intelligent agents are specific instances of AI Agents. Currently, most AI Agent projects on the chain belong to this category. In addition to the aforementioned AIXBT and Luna, there are some distinctive intelligent agents such as Freysa, Zerebro, and Pippin.

Freysa is a competitive proxy game, composed of an agent that controls the prize pool. The goal of the game is to persuade the agent to send the prize pool to oneself. Anyone can send a message to Freysa by paying a query fee.

Zerebro is an intelligent agent on Solana, focused on art creation and content generation, capable of independently creating music, memes, artworks, and NFTs. The Zerebro team has also developed the Python-based AI agent framework ZerePy.

Pippin is both an intelligent agent and a framework. Its image is derived from the abstract concept posted by Yohei Nakajima, the author of the well-known AGI framework BabyAGI, on Twitter. Yohei also launched the AI Agent framework Pippin for creating “self-feedback” mechanism agents.

DeFAI

DeFAI is a term created by @danielesesta, who is the founder of @HeyAnonai and an important figure in the Wonderland $TIME project during the DeFi Summer of 2021. DeFAI is the combination of decentralized finance (DeFi) and artificial intelligence (AI), aiming to simplify the complex processes of DeFi through AI technology, enabling ordinary users to easily access DeFi services. DeFAI is currently the most likely direction for the practical application of AI Agents combined with blockchain.

Griffain is an intention-based blockchain terminal designed to simplify the complexity of on-chain transactions through agent technology. Griffain also supports users in creating their own agents to complete various social platform interactions and DeFi transactions.

Orbit is a platform focused on simplifying DeFi operations and accessing market information, providing users with a smooth interaction experience. Grift is the AI Agent token launched by Orbit. Orbit is one of the most comprehensive DeFi platforms, developed by the @sphereone_ team and supported by @alliancedao.

BUZZ, developed by Hive AI, is the champion project of the Solana AI Hackathon. It utilizes AI technology to provide a natural language interface, allowing users to engage in DeFi trading and management more intuitively. BUZZ is based on a cellular architecture and the Solana Agent Kit, enabling it to automatically handle complex DeFi operations such as trading, staking, and lending through user chat commands, and can also be customized into tailored strategies based on user needs.

The Potential of DeFAI

The development of AI Agents is far from over; it just needs practical applications to draw a second growth curve. Currently, DeFAI has the biggest opportunity.

There are two reasons why AI Agents still have room for development:

1. From a macro perspective, the future is undoubtedly an AI world, and the integration of Web3 and AI is inevitable. The most direct way is for AI to improve Web3. Currently, the most pressing issue with Web3 is the high user threshold. AI Agents can guide users throughout their journey in the Web3 world, assist in creating and managing on-chain transactions, answer questions in real-time, provide support, and even complete complex intent transactions.

2. From a market perspective, the sudden halt of AI Agent narratives is more due to the “trust crisis” brought about by celebrity Meme coins. Events like Trump issuing coins and Trump’s wife issuing coins have turned the blockchain into a completely PvP game, and the AI Agent, as a utility token, has also been wrongly killed off by the market.

Therefore, there is a need for applications that allow users to truly experience the improvements AI brings to Web3, and for the market to recognize that AI Agents are not just a meme. DeFAI has become a new narrative direction for AI Agents, representing the second growth curve.

The AI Agent narrative only began at the end of last year, with a maximum market cap of only 13 billion USD, having only experienced the hype of the launch platform (Virtuals) and the framework (ai16z). For infrastructure, agents, and DeFAI, many high-quality projects are still being developed and promoted, and have yet to be fully understood by users, nor have they formed a breakout effect.

Infrastructure is engaged in the shovel business and is difficult to become the main character; the Meme attributes of the intelligent agent are too strong, making it hard for the token to be empowered; however, DeFAI can help users lower the threshold for DeFi trading and management, aligning with the direction of AI improving Web3. Moreover, the DeFAI token is easier to empower, such as through token payments or staking to unlock advanced features like custom AI models, trading strategies, quantitative robots, etc.

In the first growth curve, the AI Agent framework acts as a glue, connecting social platforms and various blockchains. However, the framework itself is a tool and is difficult to commercialize. Although there are tokens in Web3, the framework itself struggles to capture value and must develop into a platform with community attributes. If it only serves as a platform for launching agents, it returns to the pump.fun model; how can agents capture value? What is the difference between agents that cannot capture value and Memes? DeFAI is clearly the application that can most easily capture value, and the support of the AI Agent framework for DeFAI is an inevitable trend, with the possibility of new frameworks specifically supporting DeFAI applications emerging in the future.

Conclusion

In the summer of 2020, DeFi projects exploded on Ethereum, and various indicators in the cryptocurrency market significantly improved, with the overall market capitalization growing nearly 100 times. DeFi is currently the largest layer of application innovation in Web3 and one of the hottest entrepreneurial directions. We look forward to new application layer innovations in Web3 that go from 0 to 1, but often overlook the new DeFi innovations that leverage the power of AI to go from 1 to 100.

The value of DeFAI lies not only in simplifying the complexity of DeFi operations and enhancing the on-chain user experience, but also in standing at the intersection of DeFi and AI Agents, which can promote the popularization of DeFi, making it more user-friendly for both novices and experienced users, while also providing practical application scenarios for AI Agents, moving them from the narrative stage to the implementation stage.

At the beginning of the Year of the Snake in China, Trump’s tariffs triggered a significant drop in the cryptocurrency market, with the AI Agent sector being severely misjudged. As a pragmatist in cryptocurrency, I believe that practical applications are what will bring true prosperity to Web3. When the market is in panic, it is also a test of faith; believe in “the power of belief” and trust in the value of DeFAI.