During the Web3 Festival, Dr. Xiao Feng, Chairman and CEO of HashKey Group, officially released the 3rd whitepaper in the Web3 Economy research series: On-Chain Finance and Tokenization in the Era of the Agent Economy.

The whitepaper centers on the trend of AI Agents accelerating their participation in production, collaboration, and trading. It systematically explores the evolution of on-chain finance and tokenization infrastructure in the agent economy era, further outlining the developmental logic of the next phase of the Web3 economy across production, value, institutional structures, and business.

Over the past three years, HashKey Group has conducted continuous research into the Web3 economy from the two previous whitepapers.

HashKey believes that industry evolution over the last two years has validated several key judgments: Stablecoins are gradually evolving into the cash settlement layer of the on-chain financial system. RWA (Real-World Assets) are moving toward broader institutional-grade applications and blockchain has truly become an institutional tool for hosting new relations of production.

Meanwhile, as AI Agents become vital participants in a new round of production, collaboration, and exchange, the original on-chain financial system faces new upgrade requirements. This whitepaper aims to deep dive how must on-chain finance and tokenizationupgrade their infrastructure as Agents begin to participate in economic activities at scale.

The whitepaper points out that as AI Agents increasingly participate in production and trade, future economic activity will shift from human-to-human interaction to high-frequency collaboration between humans and Agents, as well as Agent-to-Agent interactions.

Traditional financial systems were not built for the Agent economy. Their logic relies on physical accounts, manual audits, intermediary brokerage, and staged clearing—processes best suited for low-frequency, high-value activities with high human involvement. Facing the high-frequency, small-amount, and automated needs of Agents, the limitations of this system in terms of efficiency, cost, and response speed become more apparent.