Ant Financial Science first proposes a new architecture for the intelligent agent economy, covering four major aspects: identity, payment, risk control, and compliance.

Title: Ant Group Digital Technology First Proposes a New Architecture for the Intelligent Agent Economy, Covering Four Dimensions: Identity, Payments, Risk Control, and Compliance

Author: Rhythm BlockBeats

Source:

Reposted: Mars Finance

On April 20, Ant Group Digital Technology first proposed an architecture concept for the intelligent agent economy—the “4R Full-Stack Architecture” (4R Full-Stack)—at the Hong Kong Web3 Festival. It covers four layers: Agentic Runtime, Payment Rails, Agent Registry, and Root Infrastructure. The aim is to provide technical infrastructure that enables AI agents to support identity, payments, risk control, and compliance.

Ant Group CTO Yan Ying said in her speech that the current foundation for the intelligent agent economy has “four cracks”: loss of execution control caused by loopholes in prompt logic; a responsibility vacuum caused by AI lacking trustworthy identities; transaction barriers caused by payment gateways being built around humans; and collaboration risks caused by unfamiliar Agents lacking trust. “This is not something that can be solved by patching software—it requires redesigning from the underlying infrastructure layer.”

According to Yan Ying, the core product of the Agentic Runtime layer is DTClaw, which is built in with the CARLI security model. At the execution layer, it enforces binding constraints on agent behavior. It supports multi-model compatibility and financial-grade compliance standards, with the goal of making every step of AI operation controllable, auditable, and recoverable.

The Payment Rails layer builds native on-chain payment channels, integrating agent intelligent decision-making with verifiable credential-chain technology. While ensuring precise identification of payment intent and full-chain security and control, it also realizes transparency and immutability across the transaction process. For high-frequency micro-transaction scenarios, the platform has constructed a native instant settlement network, supporting seamless transfers of cross-chain, multi-asset value and intelligent routing—greatly improving capital turnover efficiency. In addition, by providing a standardized development toolchain and a seamless wallet integration experience with no user friction, this approach significantly lowers development barriers and users’ usage costs, forming a payment closed loop that balances financial-grade security with an exceptionally smooth experience.

The Agent Registry layer issues on-chain identities for each agent based on DID (decentralized identity) and the ERC-8004 standard, ensuring that every collaboration between agents can be verified. The Root Infrastructure layer, as the architectural foundation, uses Jovay Layer2 to achieve 120-millisecond-level confirmations to support AI micro-payments. It also combines ZKVM technology to enable off-chain computation and on-chain verification, solving the compute-power trust issue in the AI economy. Yan Ying said, “Root Infrastructure leverages blockchain and privacy computing technologies to provide an immutable contract-execution environment for agents. Even two unknown AIs can establish trust based on code and trade with confidence.”

At present, AI is moving from dialogue (Chat) and action (Action) toward the era of the intelligent agent economy. Yan Ying believes that the qualitative shift in the third stage is not that AI becomes smarter, but that it begins to have ownership of assets and transaction rights. Yan Ying said that over the past decade or so, Ant Group Digital Technology has accumulated extensive engineering practices in financial-grade security, privacy computing, blockchain, and compliance systems. Based on this foundation, the 4R architecture has undergone new R&D and design.