Written by: WOO
Background: The AI landscape is more hype-driven than meme coins
AI has always been seen as the biggest narrative outside of meme coins in this cycle, but the truth is that most AI projects are jokingly referred to as being no different from meme coins - firstly, because Web 3 AI projects have very few actual products, and secondly, because the narrative of combining AI and Web 3 does not come naturally, often giving the impression of crypto projects forcibly piggybacking on AI.
The most successful recent case of AI being applied in crypto is its combination with meme coins, with the emergence of GOAT sparking market excitement, and a variety of AI + meme coin projects popping up like mushrooms after the rain, such as GNON, ACT, SLOP, and Shoggoth.
In fact, the aforementioned meme coins are not just simply adding AI elements, but are actually created by AI Agents. So what is an AI Agent? What are some notable projects in this area? WOO X Research will take a look in this article.
What is an AI Agent?
An AI Agent refers to an artificial intelligence that can autonomously complete tasks without human intervention.
Traditional large language models provide you with rough information and a framework, while AI Agents can provide more detailed, use-case-specific information. Therefore, AI Agents have greater potential in solving complex problems and automating tasks, and can provide more specific and practical solutions.
For example, if you ask an AI how to buy the cheapest headphones online, an LLM might provide a general answer outlining the basic steps. But an AI Agent would actually help the user place the order directly online, like a personal assistant in real life.
Current representative AI Agent products in Web 2 include the search-focused Perplexity, Google's Notebooklm for organizing long-form content, and the virtual character creation system Character, among others.
How can AI Agents be combined with Web 3?
Model training: Infrastructure-type projects need users to provide data as training material for the models, and users who provide data can receive project revenue sharing in the form of tokens. The more users, the higher the platform's revenue, which will attract more data providers, allowing the model training to be further improved. This approach has a flywheel effect compared to the centralized data model of Web 2, but so far the effect has not been significant due to the early stage of AI development and the low user base of Web 3.
Bridging information gaps within the industry: The transparency of blockchain allows the creation and use of financial products and other services without permission, but also leads to an overwhelming amount of market information scattered across different platforms and applications, making it difficult for users to obtain comprehensive and accurate information. AI Agents can integrate data from different blockchains to provide clear, simplified analysis and recommendations, helping users better understand market dynamics, project risks, and opportunities.
AI Agent tokenization: AI Agents created by users can be sold on the market, providing incentives for users to build AI Agents, while also enhancing the product richness of the AI Agent issuance platform and strengthening its ability to capture token value.
Here is a brief introduction to three AI Agent projects:
Myshell: Users create their own AI Agents
Myshell is a platform that allows users to create their own AI Agents, and users can also sell their designed AI Agents on the official marketplace to earn profits, with transactions denominated in BNB. The current market is dominated by anime-style chatbot AI Agents, covering areas such as work assistants, data lookup, and online learning, with a variety of language models available, including GPT series, Gemini, Claude, and Meta Llama.
The platform is currently running an ongoing points activity, and will launch its native token SHELL in the future, which will be used to unlock advanced features, reward bot developers, and pay usage fees.
The team has completed three rounds of financing, raising at least $16.6 million in total. The highlights include an investment from Binance Labs in August 15, and a Pre-Seed round led by the well-known crypto VC Dragonfly, with a total of 19 participating institutions and independent investors. It's worth mentioning that Balaji Srinivasan, former Coinbase CTO and a16z general partner, is also an investor.
Virtual: An AI Agent platform focused on entertainment
Virtual believes that AI Agents are the future of AI, and they focus on enabling these agents to run in applications and games, and leverage Web 3 features to tokenize and co-own these AI Agents. A simple example would be using AI Agents as NPCs in RPG games - if each NPC character is a different AI Agent that can exist across game platforms, it can create a deeper, more sticky gaming experience for players, ultimately increasing game revenue.
Virtual Protocol's VIRTUAL Agents currently have the ability to move and speak in 3D space, conduct on-chain transactions with their own wallets, and most importantly, be tokenized and traded. The representative project is Luna, an AI Vtuber that live streams 24/7 on YouTube (https://www.youtube.com/watch?v=b19rf34ffTM), and users can chat and interact with it on Virtual's website.
AI Agents created using Virtual can only be redeemed using the platform's token VIRTUAL, and different fees are charged when buying and selling, which also enhances VIRTUAL's ability to capture value. Currently, Virtual's market cap is around $110 million, and Luna's market cap is around $50 million. Other tokens like Game and Convo have also performed well recently.
Creator.Bid: A meme coin platform with AI filters
Creator.Bid focuses on providing the easiest way to create your own AI Agent, which can then be tokenized and shared on platforms like Twitter, Telegram, and CreatorBid itself to earn rewards.
Non-Fungible Tokens (NFTs) are used to ensure the uniqueness of AI Agents, guaranteeing that no two agents share the same code. These NFTs can also be bought, sold, and transferred, providing flexibility. The model is similar to meme coin launchpads, where users have to input an initial purchase amount after creating the agent. This issuance method inevitably reminds one of projects like Friend Tech and Pump Fun, and the platform will later issue a token $AGENT as a value capture and governance tool, making the overall product economic model more flywheel-like.
The project has completed financing, with the amount undisclosed. Investors include Mechanism Capital (early ARB investor), Zee Prime Capital (early SOL investor), and Moonrock Capital (early MNT investor), among others. The combination of Ponzi, SocialFi, and AI Agent concepts is worth continued attention.
