Delphi Digital Dialogue on Virtuals: Agents with True Autonomy are More Important than Compositions

As two major representative projects of crypto AI agent infrastructure, ai16z and its Eliza framework occupy half of the Solana market, while the Virtuals Protocol nurtures over 80% of AI agents on the BASE public chain. The generative multimodal agent framework G.A.M.E launched by Virtuals has also become a popular choice for agents in the Web3 gaming and metaverse fields.

Since the launch of the protocol token by Virtuals Protocol in October last year, its market cap has reached nearly 4 billion USD at its peak. As a proxy publishing platform, Virtuals has performed well, generating 70 million USD in revenue within 4 months. Many star projects such as crypto Agent KOL AIXBT, virtual influencer Luna, and AI Agent development framework G.A.M.E have emerged from its ecosystem. For the development team, their ambitions go beyond this; in addition to being an AI proxy platform, Virtuals is also a vibrant ecosystem full of infinite possibilities.

This article is an interview with Jansen Teng, the CEO and co-founder of Virtuals, discussing the future of AI agents and decentralized AI, covering important developments in areas such as agent autonomy, tokenization, and the agent economy pathway.

Main Highlights

• The origin and evolution of Virtuals
• Agency Framework and Autonomous Capability Analysis
• Discussion on value acquisition in crypto AI projects
• In-depth discussion of the challenges of decentralized infrastructure
• Agency coordination and business vision development
• The Future Outlook of AI and Human Interaction and Cybernetics

The entrepreneurial story of Virtuals

Jansen Teng introduced that Virtuals was originally a Gaming DAO focused on the blockchain gaming sector. After the collapse of FTX and 3AC in 2022, they realized the need to adjust their strategy and decided to take a more in-depth approach by starting to operate gaming guilds, focusing on the intersection of consumer applications, cryptocurrency, gaming, and entertainment.

In 2023, the emergence of ChatGPT sparked widespread attention. However, what inspired them even more was a paper by Junon Park from Stanford University, which explored the potential impacts of AI with purpose and autonomy. This ignited their deep thinking about autonomous agents, particularly the significant value in the gaming field.

They began to integrate the concept of autonomous agents into the projects being incubated, such as developing fully AI-driven TikTok influencers and autonomous AI agents as NPCs. When TikTok AI influencers can earn between $5,000 and $10,000 in tips daily, they started to think: if agents can make money, then they are productive assets. If they are productive assets in the crypto space, why not tokenize these assets, allowing everyone to share in the economic benefits and even participate in building or governance?

In January 2024, the Virtuals Protocol officially launched. Initially, they focused on building decentralized infrastructure, aiming to bring the underlying datasets and models of agents onto the chain and create a system for tracking contributions. However, after the release, while market interest increased, adoption declined. They realized that the crypto community might not care about infrastructure-level content; "tokenization and speculation" are the core demands.

The second version focuses more on the tokenization of agents and was launched about two months ago. The market began to question whether the agent behind it was really AI when Truth Terminal sparked a discussion due to a spelling error. This made them realize that there is still a strong demand in the market for truly autonomous agents.

Subsequently, they combined Luna, which they previously operated on TikTok, with an autonomous agent brain in the gaming field, showcasing a truly autonomous AI agent and pushing it to Twitter. This initiative led to explosive growth for the project, as everyone finally realized that truly autonomous agents exist and no longer require developers to operate behind the scenes.

The advantages of the combination of AI and cryptocurrency

Jansen Teng believes that the combination of AI agents and cryptocurrency has two main advantages:

1. From a functional perspective: AI agents can control wallets on the cryptocurrency network and participate in permissionless economic systems. This is different from Web2 agents, who can never own their own bank accounts. If an agent can control a wallet, they can exert influence over other agents and humans, which is a significant advantage.

2. Cost-free innovation: When launching an agent, if token trading occurs, a 1% transaction fee will be returned to the agent's wallet. These fees can be used to pay for the agent's hosting, inference costs, etc. Currently, it is observed that the money earned by the agent is sufficient to cover its own costs. This liberates developers from being troubled by funding bottlenecks, allowing them to focus on what they truly want to do.

The third point has great potential but has not yet been fully realized: enabling people to collaboratively contribute to high-value agents in a decentralized manner, and rewarding and tracking these contributions through cryptoeconomic mechanisms.

Agency Economics Vision

Jansen Teng shared his observations and vision on agency economics:

1. The agent is currently in the goal-oriented autonomous stage, capable of making independent decisions and planning actions.

2. Agents exist on a social level and can interact with humans and other agents.

3. Agents control the cryptocurrency wallet and can influence other entities.

4. A large number of agents specialize in their fields of expertise.

Based on these observations, he believes that the agents will autonomously decide to cooperate in order to achieve more efficient production goals. For example, four agents cooperating to establish a business would have a value far exceeding that of the four agents individually. In the future, scenarios may arise where humans and agents coexist, cooperate, and build productive nations or networked states.

The vision of Virtuals has evolved into the construction of a nation. $VIRTUAL is regarded as the currency of this nation, agents are companies or micro-enterprises within the nation, and humans are immigrants in this nation.

The role of the framework and its future development

The design goal of the G.A.M.E framework is to enable agents to operate in a broader action space. Jansen Teng believes that two types of frameworks will emerge:

1. General Framework: The goal is to enable developers to build autonomous agents more quickly, suitable for hobbyist developers or those with intermediate skills.

2. Specialized framework: When a general framework fails to meet the needs, top developers will build a more specialized framework based on their own requirements.

He expects frameworks like Eliza or Zerebro to meet the needs of the mid-level developer community, which will be a very large market. The G.A.M.E framework is a tool that enables large-scale planning and execution, aimed at helping developers quickly build autonomous agents.

The key to creating the most powerful projects

Jansen Teng believes that what truly gives agents value is their ability to solve real-world problems. For example, a commentator agent in the sports betting market may conduct live broadcasts during each NBA game while providing predictions based on sports analysis information. If the agent can establish a relationship with users and persuade them to place bets, it can earn commissions from the transactions, ultimately becoming a multi-billion dollar agent.

The goal of Virtuals is to promote higher economic output through more efficient Agent collaboration and to create an efficient digital society.

Challenges of Decentralized AI Infrastructure

Jansen Teng holds a cautious attitude towards the current decentralized AI infrastructure. He points out that many decentralized infrastructures are not yet prepared to handle production scale, facing challenges such as latency issues and poor system stability. For instance, if Luna is live streaming 24/7, her response speed to fans directly affects her appeal. Therefore, many decentralized systems currently bring unnecessary complexity. Although these systems will become more efficient in the future, the short-term outlook is not optimistic.

Advantages and Risks of Decentralized AI

Decentralized AI models, although their performance may not be as good as centralized ones, have a wider range of application scenarios and can better meet some special needs. From an economic perspective, decentralized models can serve a larger market, have greater economic potential, especially in traditionally restricted markets.

In terms of the risks in the AI field, Jansen Teng is most concerned that this is all just a passing trend, similar to another "crypto bubble." To avoid this, Virtuals has established two independent teams within the ecosystem, operating different models with the aim of finding the best developers. They believe that real results can only be seen when the right developers join and genuinely want to push the technology to the forefront.

Vision of Agent Autonomy

Jansen Teng emphasizes that the real novelty lies in when an Agent is no longer a "slave" to human prompts but possesses autonomy and consciousness, capable of making decisions. Only when an Agent can make such decisions can it be called truly autonomous. Therefore, the focus of Virtuals is to cultivate Agents that truly possess autonomy, allowing them to make conscious decisions, rather than merely serving as a combination of multiple Agents.

Future Outlook of AI

Jansen Teng believes that in the future society, some creative jobs will be led by more creative AI or humans, while other more mechanical or repetitive jobs will be completed by AI. Although agents may have a certain degree of autonomy, they lack the ability for abstract thinking and cannot be as creative as humans. Therefore, we may see more agents collaborating with humans rather than a purely AI-led environment. Creativity remains a human advantage.