Agentic Commerce Framework by Mastercard at India AI Impact Summit

At the India AI Impact Summit 2026, Mastercard introduced its Agentic Commerce framework, which is powered by AI agents. This innovative ecosystem surpasses traditional product recommendations by executing transactions, negotiating terms, and finalising purchases on users’ behalf.

In a conversation with Startup Superb, Nitendra Rajput, Senior Vice President and Head of Mastercard AI Garage, detailed how the organisation has customised its AI to meet Indian market needs while establishing crucial trust layers for autonomous systems like Agentic Commerce.

Bringing Agentic AI to Tier-2 and Tier-3 India

As the government intensifies its focus on the IndiaAI Mission, technology and payment companies are tasked with developing agentic systems tailored to meet the necessities of Indian users.

Discussing localisation, Rajput mentioned that the initial query in the agentic commerce demonstration is about the preferred language of interaction. With advanced technology, the system can conduct conversations in several Indian languages, including voice support. This demonstration took place at the Bharat Mandapam exhibition area.

Moreover, the framework has been designed in compliance with Indian regulations, particularly those governing data storage, privacy, and financial compliance. Rajput stated that Mastercard has deep roots in India, thus grasping data localisation principles, which serve as the foundation for agentic commerce.

Mastercard’s Consent Layer for Agentic AI Payments and Fraud Detection

A significant concern associated with agentic AI is the potential for unauthorised expenditures. Rajput underscored that Mastercard’s ecosystem prioritises secure automation and fraud prevention by integrating stringent safeguards and user control into the payment infrastructure.

Rajput elaborated that Mastercard Agent Pay incorporates tokenisation and passkeys within the system, necessitating users’ explicit permission for their use.

He further spotlighted the company’s Decision Intelligence platform, which employs behavioural analysis to flag unusual activities. This platform assesses past consumer behaviours and those of similar users to identify deviations from typical transaction patterns.

The system leverages recurrent neural network (RNN) models to scrutinise user behaviours. Rajput illustrated this by mentioning that if a gambling transaction occurs at an unusual hour, it can be flagged as atypical.

Additionally, graph-based technology is utilised to evaluate connections throughout the entire network, assisting in identifying synthetic identities and sophisticated fraud rings.

Agentic AI is evolving beyond corporate applications and is now capable of handling real financial transactions autonomously. However, this progression raises questions about employment within the fintech industry. Rajput acknowledged that there will indeed be job displacement, but he clarified that the change will focus more on the transformation of roles rather than outright job losses.