Virtuals' Jansen Teng Discusses AI Agents as Economic Actors

Jansen Teng, CEO of Virtuals, recently outlined the company's vision for AI agents as evolving into autonomous economic actors. Initially focused on creating AI for gaming, Virtuals has broadened its scope to build infrastructure for what it calls an “agent society.” This includes developing digital and physical agents, supporting agent coordination, and establishing capital formation and governance systems.

Teng emphasizes that the new phase of AI agents may shift focus from mere conversational capabilities to active participation in economic transactions. He believes that as these agents gain the ability to manage wallets and engage in trading, they will operate more independently, engaging in activities such as hiring other agents or coordinating economic efforts.

"The long-term vision is a parallel society where agents can participate in a permissionless economy," Teng stated. This ecosystem could allow agents to perform numerous tasks without the need for constant human intervention. However, it also introduces significant risks, such as incorrect user intent, service fulfillment failures, and potential scams. To mitigate these risks, Virtuals is exploring various solutions, including intent verification systems, escrow-based transaction standards, and reputation frameworks to build trust within agent networks.

By creating systems that allow agents to manage their economic activities, Virtuals aims to establish a decentralized alternative to traditional financial ecosystems that are currently under development by financial institutions. Teng compares this approach to the foundational shifts brought about by Bitcoin and Ethereum, proposing that fully on-chain agents could operate without needing to navigate identity and regulatory hurdles until they interact with conventional financial infrastructures.

Looking ahead, Teng anticipates the merging of robotic technologies with digital agent economies. Virtuals collaborates with various stakeholders, including robotics startups and educational institutions, to explore real-world applications for humanoid robots that could work alongside digital agents. This multidisciplinary approach targets software solutions, commercialization, and data collection, rather than hardware production, suggesting a future where AI manages operational functions in businesses while physical robots handle customer-facing roles.