The demonstrations showed how consumers can use natural-language prompts to search and discover products and services, compare options and proceed towards purchase through an AI agent. The use cases included travel bookings, food ordering, lifestyle purchases and online shopping.
The demonstrations were carried out with a group of banks, fintech companies and merchant and payment technology platforms, including Federal Bank, IndusInd Bank, RBL Bank, Cashfree Payments, CCAvenue, Juspay, Paytm Payments Services, PayU, Pine Labs and Razorpay.
How agentic commerce works
In a conventional online transaction, a consumer typically searches for a product or service, evaluates alternatives, selects an option and then completes the payment.
Agentic commerce seeks to bring these steps into an AI-assisted interaction. Visa’s demonstration showed consumers starting with a natural-language request, either through a popular large language model (LLM) AI agent or directly through a merchant’s AI agent.
The AI can then assist with discovery and comparison before the transaction moves to payment.
The payment layer is critical because the AI-assisted journey still needs to preserve authentication, consumer consent and control. Visa said its demonstrations used existing payment capabilities, including tokenisation and Visa Payment Passkey-based authentication.
The company said the showcased transactions involved fully authenticated payments with a human-in-the-loop, rather than removing the consumer entirely from the payment process.
What it could mean for consumers
If agentic commerce scales, the way consumers interact with merchants could change significantly. Instead of navigating multiple websites or apps to search, compare and buy, consumers could increasingly use an AI interface to initiate and manage parts of the commerce journey.
That also puts greater emphasis on how payment authorisation works when an AI agent is involved.
Visa’s showcase focused on maintaining transparency and consumer control while enabling AI agents to assist with transactions. Its partner ecosystem at GFF included banks, fintechs, payment platforms and merchants, highlighting the multiple layers that would need to work together for such transactions.
The company said the demonstrations are intended to show how existing payment infrastructure, including tokenisation and secure authentication, can support emerging AI-driven commerce models.
Agentic commerce could become a large payments market
The potential market is significant. Visa cited a McKinsey estimate that AI agents could facilitate $3 trillion-$5 trillion in global commerce transaction value by 2030.
For the payments industry, this could create a new layer between consumers and merchants, with AI agents increasingly involved in product discovery, decision-making and the transaction process.
Visa’s GFF 2026 showcase, however, is a demonstration of these payment experiences rather than an announcement that consumers can now broadly hand over shopping and payment decisions to AI agents.
The company said it plans to work with banks, fintechs and merchants as the ecosystem develops, with a focus on secure, interoperable and scalable payment experiences.
The bigger shift could therefore be less about replacing existing digital payment methods and more about changing the interface through which consumers discover, decide and ultimately pay for what they buy.
