Avichal Garg, a partner at Electric Capital, points out that as AI agents become increasingly autonomous, developers are starting to configure them with crypto wallets, enabling the software to hold assets, pay service fees, trade tokens, and even hire other AI agents. This trend is driving crypto technology into a new phase—building financial systems for “non-human entities,” but the relevant legal framework is still significantly lagging behind. He believes that with the programmable funds, instant settlement, and global accessibility of blockchain, AI agents can not only make decisions but also independently complete transactions, thus forming software entities that can “think and execute financial activities.” Garg states that this model is similar to the emergence of the limited liability company system in the 19th century, releasing new productivity thresholds for economic activity. As the cost of participation continues to decrease, more individuals and teams worldwide can create economic value with the help of AI agents. However, the core issue remains the definition of legal liability. Since AI itself cannot be punished, there is still no clear answer as to who should bear the responsibility if an AI agent with an independent wallet participates in transactions, lending, or business activities and causes losses. This issue may become a fundamental problem that regulators must face in the future.