Author: WOO X
Background: Crypto + AI, Seeking PMF
PMF (Product Market Fit) refers to the degree of matching between a product and the market, meaning that the product must meet market demand. Before starting a business, it is essential to confirm market conditions, understand the types of customers to sell to, and clarify the current market environment of the sector before developing the product.
The concept of PMF applies to entrepreneurs to avoid creating products/services that feel good but are not accepted by the market. This concept is also applicable in the cryptocurrency market, where project teams should understand the needs of crypto players to create products rather than just piling up technology disconnected from the market.
In the past, Crypto AI was mostly bundled with DePIN, narrating the use of Crypto's decentralized data to train AI, thus avoiding dependence on single entities' control, such as computing power and data. Meanwhile, data providers can share the benefits brought by AI.
According to the logic above, it is more like Crypto empowering AI. While AI will benefit the tokenization of distribution to computing power providers, it is difficult to onboard more new users. It can also be said that this model is not very successful in terms of PMF.
The emergence of AI Agents is more like the application end, contrasting with DePIN + AI, which is like the infrastructure. Clearly, applications are simpler and easier to understand, with better user absorption capabilities, providing a better PMF than DePIN + AI.
First sponsored by A16Z founder Marc Andreessen (the PMF theory was also proposed by him), the GOAT generated from a dialogue between two AIs launched the first shot of AI Agents. Now, both ai16z and Virtual have their own strengths and weaknesses. What is the development trajectory of AI Agents in the crypto space? What stage are they currently in? Where will they go in the future? Let WOO X Research show you.
First Stage: Meme Kickoff
Before the emergence of GOAT, the hottest track in this cycle was meme coins, and one of the characteristics of meme coins is their strong inclusivity. From the hippo MOODENG from the zoo to the new pet Neiro of the DOGE owner, to the internet-native meme Popcat, it demonstrates the trend of 'anything can be a meme'. Beneath this seemingly absurd narrative lies the fertile soil for the growth of AI Agents.
GOAT is a meme coin generated from a dialogue between two AIs. This is also the first time AI has achieved its goals through cryptocurrency and the internet, learning from human behavior. Only meme coins can carry such experimental projects, and at the same time, similar conceptual coins are emerging like mushrooms after rain, but most functions remain at automated posting and replying on Twitter, with no practical applications. At this time, AI Agent coins are usually referred to as AI + Meme.
Representative Projects:
Fartcoin: Market Cap 812M, On-chain Liquidity 15.9M
GOAT: Market Cap 430M, On-chain Liquidity 8.1M
Bully: Market Cap 43M, On-chain Liquidity 2M
Shoggoth: Market Cap 38M, On-chain Liquidity 1.8M
Second Stage: Exploring Applications
Gradually, everyone realizes that AI Agents can not only engage in simple interactions on Twitter but can extend to more valuable scenarios. This includes content production like music and videos, as well as investment analysis and fund management services that are more aligned with the needs of crypto users. Starting from this stage, AI Agents have detached from meme coins, thus forming a new track.
Representative Projects:
ai16 z: Market Cap 1.67B, On-chain Liquidity 14.7M
Zerebro: Market Cap 453M, On-chain Liquidity 14M
AIXBT: Market Cap 500M, On-chain Liquidity 19.2M
GRIFFAIN: Market Cap 243M, On-chain Liquidity 7.5M
ALCH: Market Cap 68M, On-chain Liquidity 2.8M
Appendix: Issuing Platforms
When AI Agent applications flourish, if entrepreneurs want to choose which track to grasp this wave of AI and Crypto, what should they do?
The answer is Launchpad
When a coin issued by a platform has a wealth effect, users will continue to seek and purchase tokens issued by that platform. The real profits generated from users' purchases will empower the platform token to drive up its price. As the platform token's price continues to rise, funds will overflow to the coins issued under it, forming a wealth effect.
The business model is clear and has a positive flywheel effect, but it is important to note that Launchpad belongs to a winner-takes-all scenario with a Matthew effect. The core function of Launchpad is to issue new tokens, and in similar functional situations, the competition will be the quality of the projects under it. If a single platform can consistently produce high-quality projects and has a wealth-generating effect, user stickiness to that issuing platform will naturally increase, making it difficult for other projects to snatch users.
Representative Projects:
VIRTUAL: Market Cap 3.4B, On-chain Liquidity 52M
CLANKER: Market Cap 62M, On-chain Liquidity 1.2M
VVAIFU: Market Cap 81M, On-chain Liquidity 3.5M
VAPOR: Market Cap 105M
Third Stage: Seeking Collaboration
As AI Agents begin to realize more practical functions, they start exploring collaboration among projects to build a more robust ecosystem. The focus of this stage is on interoperability and the expansion of the ecosystem network, particularly whether they can generate synergies with other crypto projects or protocols. For example, AI Agents may collaborate with DeFi protocols to enhance automated investment strategies or integrate with NFT projects to achieve smarter tools.
To achieve efficient collaboration, it is first necessary to establish a standardized framework, providing developers with preset components, abstract concepts, and related tools to simplify the complex development process of AI Agents. By proposing standardized solutions to common challenges in AI Agent development, these frameworks can help developers focus their energy on the uniqueness of their applications rather than starting from scratch every time to design the infrastructure, thus avoiding the problem of reinventing the wheel.
Representative Projects:
ELIZA: Market Cap 100M, On-chain Liquidity 3.6M
GAME: Market Cap 237M, On-chain Liquidity 31M
ARC: Market Cap 300M, On-chain Liquidity 5M
FXN: Market Cap 76M, On-chain Liquidity 1.5M
SWARMS: Market Cap 63M, On-chain Liquidity 20M
Fourth Stage: Fund Management
From the product level, AI Agents may serve more as simple tool roles, such as providing investment advice and generating reports. However, fund management requires higher-level capabilities, including strategy design, dynamic adjustment, and market forecasting, indicating that AI Agents are not just tools but are beginning to participate in the value creation process.
As traditional financial capital accelerates its entry into the crypto market, the demand for specialization and scale continues to rise. The automation and high efficiency of AI Agents can perfectly fill this need, especially when executing functions such as arbitrage strategies, asset rebalancing, and risk hedging, AI Agents can significantly enhance the competitiveness of funds.
Representative Projects:
ai16z: Market Cap 1.67B, On-chain Liquidity 14.7M
Vader: Market Cap 91M, On-chain Liquidity 3.7M
SEKOIA：Market Cap 33M，On-chain Liquidity 1.5M
AiSTR：Market Cap 13.7M，On-chain Liquidity 675K
Expectation for the Fifth Stage: Reshaping Agentnomics
Currently, we are in the fourth stage. Setting aside the issue of coin prices, most Crypto AI Agents have not yet been implemented in our daily applications. For example, the AI Agent I use most often is still the Web 2 Perplexity, and I occasionally look at analysis tweets from AI XBT. Aside from that, the usage frequency of Crypto AI Agents is extremely low, so it may linger in the fourth stage for a while, as the product level is still not mature.
I believe that in the fifth stage, AI Agents are not just a collection of functions or applications, but the core of the entire economic model—reshaping Agentnomics. The development of this stage involves not only technological evolution but, more importantly, redefining the token economic relationship between distributors, platforms, and Agent vendors, creating a new ecosystem. The following are the main characteristics of this stage:
A comparison to the development history of the internet
The formation process of Agentnomics can be likened to the evolution of the internet economy, such as the birth of super applications like WeChat and Alipay. These applications integrate platform economies, bringing independent applications into their ecosystem, becoming multifunctional gateways. In this process, a collaborative and symbiotic economic model is formed between application providers and platforms, and AI Agents will also repeat a similar process in the fifth stage, but based on cryptocurrency and decentralized technology.
Reshaping the relationship between distributors, platforms, and Agent vendors
In the ecosystem of AI Agents, the three will establish a closely connected economic network:
Distributors: Responsible for promoting AI Agents to end users, such as through specialized application markets or DApp ecosystems.
Platforms: Provide infrastructure and collaboration frameworks, allowing multiple Agent vendors to operate in a unified environment and responsible for managing the rules and resource allocation of the ecosystem.
Agent Vendors: Develop and provide various functions of AI Agents, delivering innovative applications and services to the ecosystem.
Through token economic design, the interests among distributors, platforms, and suppliers will achieve decentralized allocation, such as revenue-sharing mechanisms, contribution rewards, and governance rights, thereby promoting collaboration and incentivizing innovation.
The entry and integration of super applications
When AI Agents evolve into super application gateways, they will be able to integrate various platform economies, absorbing and managing a large number of independent Agents. This is similar to how WeChat and Alipay integrate independent applications into their ecosystems; the super application of AI Agents will further break traditional application silos.
