India’s National Payments Corporation of India is developing a Unified Agent Protocol that would let verified AI agents initiate UPI transactions on behalf of users, according to a Business Standard report. The protocol is being framed as a trust and verification layer on top of the existing UPI system rather than a replacement for it.
The proposed framework would allow AI agents to be registered, verified and authorized before they interact with UPI. It is designed to create a common protocol that can work across the ecosystem without changing the underlying payments rails.
The move builds on pilot projects already underway in India’s agentic payments space. Razorpay and NPCI launched a pilot with OpenAI in October 2025 for agentic payments on ChatGPT, while Razorpay and NPCI later announced agentic payments on Anthropic’s Claude at the India AI Impact Summit in February 2026. Those pilots used UPI Reserve Pay and were positioned as controlled, consent-based payment experiences.
The broader context is a fast-growing UPI network. NPCI data showed UPI processed 23.2 billion transactions in May 2026, before easing slightly in June. NPCI has also been exploring agent-to-agent workflows to speed up compliance cycles, signaling that the new protocol could support both payments and operational automation.
Read Article: Anthropic Unveils Reflect to Help Claude Users Better Understand Their AI Habits
