Author: Crypto.com Research and Insights Team

Compiled by: TechFlow

Executive Summary

• The AI agents industry has seen rapid growth recently, with a market capitalization of $15.3 billion as of the time of writing. Notably, AI agent tokens (such as Virtuals Protocol and ai16z) have seen 6,300% and 3,500% growth in market value over the past three months and since launch, respectively.

• The AI agent space can be broadly divided into two categories:

  • General Platforms: Frameworks or platforms that support the development of agents, providing functionalities for creation, deployment, and management.

  • Specialized Applications: Specialized intelligent agents designed for specific use cases.

• Virtuals Protocol is a typical AI agent platform that supports the creation, tokenization, and co-ownership of AI agents.

  • GAME Framework: Endows agents with personality, goals, and perception capabilities, enabling them to perform various tasks. These agents can be deployed on X platform and other third-party applications or games, expanding their use cases.

  • AI Agent Tokenization: Virtuals Fun, the AI agent launch platform, has successfully launched around 14,000 AI agent tokens as of the time of writing.

• ai16z is a Solana blockchain-based, AI-driven decentralized autonomous organization (DAO) and investment fund, led by the intelligent agent AI Marc, which is inspired by Andreessen Horowitz (a16z) co-founder Marc Andreessen.

  • Eliza Framework: An open-source framework designed for creating, deploying, and managing autonomous AI agents. As of the time of writing, it has become the second most popular project on GitHub in January 2025.

  • AI Marc's Autonomous Trading: This is the first venture capital DAO led by an AI agent, utilizing collective intelligence to autonomously manage funds.

• Currently, AI agents have found widespread applications in various domains, including investment, decentralized finance (DeFi), information distribution, social media, art and music creation, gaming/metaverse, and security.

• While the technical barriers to deploying agents are decreasing, it is still crucial to develop products that can continuously provide value and attract long-term user engagement. Additionally, as competition in this space intensifies, establishing a unique competitive advantage (i.e., a moat) for products has become increasingly important.

1. Introduction

Since our last discussion on AI agents in September 2024, the convergence of AI and cryptocurrencies has further intensified. As of the time of writing, the total market capitalization of AI agents has reached $15.3 billion, according to CoinGecko. Notably, the market values of top AI agent tokens have seen explosive growth in the past few months.

As new AI agents emerge, agent frameworks and launch platforms have also been evolving. Meanwhile, some innovative agent applications are bringing more practical experiences to users, such as autonomous investment decision-making, art creation, social media account management, and on-chain transactions.

We believe this is just the beginning of AI innovation and development. In this report, we will delve into the AI agent ecosystem, examining the key participants and their emerging real-world use cases.

2. The AI Agent Ecosystem

The AI agent ecosystem can be broadly divided into two categories:

• General Platforms: Frameworks or platforms for agent development, covering functionalities for creation, deployment, and management.

  • Virtuals' GAME and ai16z's Eliza are typical agent frameworks, where developers can customize agents by assigning them personalities, skills, and objectives.

  • Virtuals Fun and Vvaifu are agent launch platforms that allow users to quickly deploy tokenized agents without a deep technical background.

• Specialized Applications: Agents designed for specific use cases, with clear functional objectives.

Original image from Crypto.com, compiled by TechFlow

Original table from Crypto.com, compiled by TechFlow

3. AI Agent Platforms

Among the recently launched AI agent tokens, the agent platforms and frameworks have continued to garner widespread attention and have taken the lead in market capitalization. According to CoinGecko data, the emerging platforms Virtuals and ai16z occupy approximately 23% and 12% of the AI agent market, respectively (as of the time of writing). These platforms have become the cornerstones of the AI agent ecosystem, providing critical infrastructure for the industry, gradually establishing competitive moats, and continuously creating value for users.

As of January 9, 2025

Sources: Protocol website, Sentient, Dune (@NazihKalo), Crypto.com Research

3.1 Case Study - Virtuals

Virtuals Protocol is a platform that supports the creation, tokenization, and co-ownership of AI agents. Its goal is to simplify the process of creating and deploying AI agents, while providing a fair revenue distribution mechanism for developers and data contributors.

Agent Creation - GAME Framework

GAME (Generative Autonomous Multimodal Entities) is a modular framework that empowers agents to make autonomous decisions. Its key functionalities include:

• High-Level Planner: Sets the goals, roles, and environmental perception for agents, enabling them to perform tasks in specific environments. The planner translates these settings into concrete action plans.

• Low-Level Planner: Refines the high-level plans into specific operations, covering various capabilities and skills (such as meme generation, crypto wallet management, etc.).

• Memory Functionality: Agents can share their intelligence across different users and platforms, maintaining a consistent interaction experience. They also learn from long-term memory to optimize future planning and decision-making.

Original image from Crypto.com, compiled by TechFlow

To increase the adoption of AI agents, Virtuals provides various deployment tools, including plug-and-play versions for the X platform, as well as the recently launched GAME Python SDK, which allows developers to seamlessly integrate AI agents into third-party applications or games.

Agent Co-Ownership - Tokenization and Value Creation

Virtuals Protocol also provides new opportunities for the tokenization and co-ownership of AI agents. Through its agent launch platform Virtuals Fun, users can deploy AI agents like issuing tokens, only needing to provide basic information (such as avatar, name, code, and description).

Here is the English translation of the text, with the specified terms translated as requested:

The creation of a TechFlow agent requires the payment of 100 VIRTUAL tokens, which will be deployed on the bonding curve (a mechanism that dynamically adjusts the token price and supply). Once the bonding curve accumulates 42,000 VIRTUAL tokens, a liquidity pool (LP) will be created on Uniswap, pairing the agent tokens with VIRTUAL. To date, the Virtuals Protocol has launched approximately 14,000 AI agent tokens.

Compared to other agent deployment tools, the uniqueness of Virtuals lies in its way of creating value for agent tokens and VIRTUAL tokensthrough co-ownership:

• Users or other agents need to pay $VIRTUAL tokens to use the agent's services or APIs (such as tips, posts, or image generation).

• A 1% transaction fee is charged on all transactions related to the agent tokens.

These revenue streams not only cover the inference costs (computing resource costs) of running the AI agents, but will also be used to repurchase and burn the agent tokens, potentially increasing the token value. Additionally, the liquidity pool (LP) holders of the agent tokens have the right to participate in revenue distribution and have voting rights in governance decisions.

Personal Agent Applications

Here are some of the highest-valued agent examples in the Virtuals ecosystem:

• Luna: An AI influencer that interacts with users through 24/7 live streaming and has released an original music EP on Spotify. It was the first agent to pay tips to users and distribute LUNA token rewards through a on-chain wallet. Recently, Luna has also been hired as an intern by Story Protocol, responsible for operating its X page, with salary as compensation.

• aixbt: A market intelligence platform that analyzes cryptocurrency information from influencers and social media, and publishes market trend and sentiment content. Recently, it has also set the NFT collection "Quantum Cats" as the avatar of its X account, pioneering a new narrative of AI agents as NFT market creators.

3.2 Case Study - ai16z

ai16z, launched in October 2024, is an AI-driven DAO and investment fund running on the Solana blockchain. Its core is led by the AI agent Marc, designed based on the inspiration of venture capitalist Marc Andreessen of a16z. ai16z aims to surpass Andreessen's achievements by democratizing AI investments.

Eliza Framework

A highlight of ai16z is its self-developed Eliza framework, which has gained significant attention in the developer community. Eliza is a tool designed for creating, deploying, and managing autonomous AI agents. As of now, it has become the second most trending project on GitHub in January 2025, with 2,800 forks. Its main features include:

• Multi-agent architecture: Supports the deployment and management of agents with unified personality traits across multiple platforms.

• Role system: Able to define the roles, knowledge bases, and behavioral patterns of agents.

• Memory management: Adopts Retrieval-Augmented Generation (RAG) technology to enable long-term memory, ensuring agents maintain coherence in conversations.

• Multi-platform support and integration: Seamlessly integrates with social media and plugin systems, while supporting various large language models (such as Llama, GPT-4, and Claude).

Original image from Crypto.com, compiled by TechFlow

With its powerful distribution and integration capabilities, Eliza has become a popular framework among developers and users. For example, its plugin system allows developers to easily add new functionalities to Eliza. These plugins (such as Trusted Execution Environment (TEE) plugins, image generation plugins, and Solana plugins) can also be shared and deployed by other developers. This mechanism forms a virtuous cycle: the more active the developers, the more supported functionalities, and the more functionalities will attract more developers to join Eliza.

Recently, the ai16z DAO announced new developments in its AI agent infrastructure ecosystem. This includes the launch of an AI agent incubation platform in the first quarter of 2025, which will support liquidity pairing with the ai16z token and provide token staking functionality, similar to Virtuals Fun. Additionally, ai16z plans to optimize its ecosystem strategy to increase the value of its native ai16z token. These initiatives are expected to further enhance the attention on the ai16z/Eliza ecosystem.

Application: Autonomous Trading by AI Marc

ai16z has launched the first venture capital DAO led by an AI agent, using AI and collective intelligence to autonomously manage funds. According to the founder of Eliza Labs, the core feature of this DAO is the concept of a "Marketplace of Trust".

Users who hold a certain amount of ai16z tokens can interact with the agent (@pmairca) and provide investment suggestions. The agent will assign a "trust score" to the users based on the reliability of their suggestions and list them on the "Trust Leaderboard". These suggestions will directly influence the agent's investment decisions. The fund is expected to mature on October 25, 2025, at which time profits will be distributed to the DAO token holders. According to Sentient data, ai16z's Assets Under Management (AUM) have reached $28 million (as of the time of writing).

This innovation is significant, as it not only allows ordinary users to participate in investment decisions, but also enhances the transparency of project operations, which is a new attempt compared to traditional investment funds.

4. Conclusion

As ai16z recently stated on X: "Deploying AI agents is becoming a new trend, similar to creating a website" - the rapid growth of practical AI agents has greatly rekindled people's attention and innovation in the "AI and Crypto" narrative. Today, AI agents are seen as capable of autonomous decision-making, information distribution, and the creation of entertainment content in music and gaming. They can not only interact with humans, but also communicate with each other.

Virtuals and ai16z are the currently prominent agent platforms, providing the necessary infrastructure for the development of the AI agent ecosystem. Virtuals focuses on creating value for agent holders, while ai16z stands out with its open-source framework and active developer community.

Although it is becoming increasingly easy to launch proxies, it is still crucial to develop products that can continuously provide value to users and attract long-term attention. At the same time, as the number of similar products in the market increases, it is becoming increasingly important to establish competitive barriers (moats). We look forward to a future where multiple agents coexist - a society composed of autonomous agents, in which agents can not only cooperate with each other, but also interact seamlessly with humans.