On April 29, 2026, OKX introduced the Agent Payments Protocol, a groundbreaking system designed to allow artificial intelligence (AI) to manage various business tasks autonomously, including negotiations, payments, and dispute resolution, eliminating the need for human intervention. The announcement was made on the exchange’s official learning page, where CEO Star Xu emphasized that this launch represents a pivotal step toward actualizing the Agent economy in real-world operations.
The new protocol allows AI agents to perform a wide range of tasks, such as browsing the internet, reviewing documents, or booking flights, all without human oversight. Similar to how one might ask an AI assistant to draft an email, OKX envisions a future where AI can autonomously handle financial transactions.
The Agent Payments Protocol is designed to revolutionize transaction processes by facilitating automated, full-cycle transactions. This encompasses price negotiations, secure escrow of funds, verification of task completion, and unconditional payment releases when specified conditions are satisfied, all executed by AI agents from initiation to conclusion.
In terms of functionality, the Agent Payments Protocol enables AI programs to oversee the entire business ecosystem, from creating quotes and negotiating terms to holding funds in escrow and resolving any disputes. This level of comprehensive management distinguishes OKX’s offering from those of competitors like Coinbase and Stripe. Coinbase’s x402 protocol, which is based on an older Internet protocol, has processed over 100 million transactions since its debut, emphasizing quick, low-cost payments. Meanwhile, Stripe has also rolled out its own innovative Machine Payments Protocol, signaling a broader recognition among major financial institutions of the potential for AI-driven transactions.
However, OKX asserts that existing solutions like Coinbase’s and Stripe’s offer limited functionality compared to its Agent Payments Protocol, which addresses the entire transaction process rather than merely executing single payments. The exchange highlighted that its protocol manages the full relationship between parties involved in a transaction, which can substantially streamline business operations.
Support for this initiative comes from a range of big tech players, including Amazon Web Services (AWS), Alibaba Cloud, and Uniswap, as well as blockchain entities like the Ethereum Foundation and Solana team. These partnerships are expected to onboard new users and developers onto the OKX platform, enabling AI integration across different services. Notably, the OKX Agentic Wallet, launched on March 18, 2026, further facilitates this ecosystem by allowing AI programs to manage their own funds securely using cutting-edge technology, such as a Trusted Execution Environment (TEE) to safeguard access.
Despite the ambitious nature of the Agent Payments Protocol, not all features are fully operational at this stage. While certain functions like one-time and batch payments, along with connections for AI-to-AI communication, are currently functional, essential components like escrow payment systems and dispute resolution mechanisms are still under development. These critical features are anticipated to play a crucial role in enhancing the efficacy of transactional business engagements.
Current trends in the AI payments marketplace indicate a nascent stage of development, as evidenced by the stark decline in daily transactions involving Coinbase’s x402, which dropped significantly from 731,000 in December 2025 to approximately 57,000 by March 2026. This represents a staggering 92% decrease, underscoring the fact that the systems in place may not yet be effectively utilized for substantial transactions. Research by Gartner suggests that over 40% of AI agent projects may be discontinued by 2027 due to rising costs and ambiguous business value.
Although these developments suggest uncertainties ahead, OKX’s ambitious initiative is reflective of a broader industry trend that anticipates a future where AI programs will increasingly take on roles traditionally filled by humans within commercial frameworks.
