By 
India is preparing to introduce an agentic payments framework on UPI, potentially allowing AI agents to execute small-value transactions on behalf of users without requiring approval for every individual payment.
The proposed system would make India one of the first countries to build agentic payments directly into a national real-time payments infrastructure. The framework, known as the Unified Agent Protocol, is expected to be unveiled at the Global Fintech Fest in Mumbai, according to people familiar with the plans.
UPI Could Become an Infrastructure Layer for AI Agents
UPI already operates at enormous scale. The system processed 24.51 billion transactions worth ₹29.82 trillion ($314.21 billion) in August 2026, according to the figures reported by Reuters.
The proposed agentic layer would allow users to give AI systems conditional authority to make payments within predefined limits.
Imagine telling an AI assistant to purchase groceries whenever a preferred product falls below a certain price. Instead of asking for confirmation at checkout, the agent could complete the transaction automatically, provided it stays within the user’s rules.
Early applications are expected to focus on low-value, frequent purchases such as groceries and ecommerce transactions. More sophisticated applications could eventually involve automated purchases based on discounts or even investment decisions triggered by predetermined price conditions.
Existing UPI Infrastructure Could Make the Transition Easier
India isn’t starting from scratch.
The proposed framework is expected to use existing UPI mechanisms such as UPI Circle, which allows a primary account holder to delegate payment authority to another user, and Reserve Pay, which allows funds to be blocked for multiple debits.
Under the proposed system, users could establish rules specifying when an AI agent is permitted to pay, how much it can spend and which merchants or services it can access.
That could make the difference between an AI assistant that simply recommends products and one that can actually complete tasks.
But giving software permission to spend money introduces obvious risks. The proposed framework is expected to include spending limits, audit trails, identity verification and merchant integration controls. NPCI is also expected to develop a liability framework covering transactions made by agents.
A New Competition in AI-Powered Commerce
India isn’t alone in exploring this model. Mastercard and Visa are developing their own agentic payment capabilities, while Indian fintech company Pine Labs has already introduced an agentic protocol called P3P.
Mastercard completed its first authenticated agentic transaction in New Delhi in June, showing that the technology is already moving beyond experimentation.
The significance of India’s approach is scale. If agentic payments are built directly into UPI, developers and merchants could potentially gain access to a massive existing payment network rather than having to create a new transaction system.
The bigger question is whether consumers will actually be comfortable giving AI systems permission to spend their money.
That trust will probably determine how quickly the technology moves from controlled use cases to everyday commerce. For now, India’s planned UPI framework suggests the next phase of digital payments may not simply involve people paying faster — it could involve AI deciding when and how those payments should happen within rules set by the user.
Comments