Mastercard has collaborated with Majid Al Futtaim to execute the first live agentic AI payment, signalling a shift from basic conversational chatbots to functional, intent-based commerce. The demonstration, conducted in Dubai, showcased a new protocol designed to bridge the gap between current consumer search habits and fully autonomous financial transactions.
Following significant engagement and a surge of industry questions via a linkedin post, regarding a live
demonstration in Dubai, Mark Walker, Editorial Director at The Fintech Times, sat down with Prakriti Singh, who is responsible for core products at Mastercard for the Eastern Europe and Eastern Africa region, to unpack the technology behind the world’s first live agentic AI payment. The interview serves as a direct response to the global interest generated by the event, which showcased a shift from basic conversational chatbots to functional, intent-based commerce in collaboration with Majid Al Futtaim. While AI agents are already used for search and comparison through large language models (LLMs), this development integrates a new protocol directly into the agentic environment to facilitate secure transactions.
Singh explained that the demonstration focused on agent-assisted commerce, a model where the consumer remains active in the process by providing real-time approval. This distinguishes it from fully autonomous agents, which are designed to make purchases independently based on pre-set parameters, such as price drops, while a user is offline. The core of this advancement is the AgentPay framework, a payment protocol that allows third-party agents to shop on behalf of a consumer. To maintain security, the system relies on three pillars: authentication of the individual, data transmission regarding the transaction’s intent, and visibility for both the bank and the merchant.
The framework utilises the Mastercard Digital Enablement Service (MDES) to replace sensitive card numbers with secure digital tokens. These tokens are bound to specific biometric data, such as facial recognition or passkeys stored on the consumer’s device. This ensures the FI can verify that the cardholder authorised the agent to act on their behalf. To manage the risks associated with third-party software, Mastercard is introducing a ‘Know Your Agent’ (KYA) process. This new standard registers agents and assigns them a unique ID, allowing issuers and merchants to identify which agent is attempting a transaction and providing them with the option to decline requests from unregistered or high-risk sources.
A significant shift in the agentic era is the reliance on ‘intent data’ for managing disputes and chargebacks. During the transaction process, the agent captures the consumer’s explicit instructions, such as the specific item and price, which is then passed through the network to the issuer. If an agent attempts to charge an amount that does not match the captured intent, the FI can flag the transaction as high-risk and decline it. This framework leverages existing debit and credit rails, ensuring that consumer protections remain intact while adapting to new methods of interaction. As merchants evaluate their strategies for LLM integration, the standardisation of these protocols is intended to maintain trust across the digital economy.
WATCH THE FULL INTERVIEW BELOW