Visa outlines trust architecture for AI agents to make purchases
Visa India & South Asia head Suresh Sethi says agentic commerce will depend on verified agent identities, authentication, tokenisation, biometrics and cyber resilience—signalling the payments infrastructure retailers may need as AI shifts from product discovery to checkout.
What happened
Visa India & South Asia head Suresh Sethi discusses agentic commerce, outlining trust, agent identity, authentication, tokenisation, biometrics and cyber
Why this matters
Payments, identity, biometrics and cybersecurity providers with agent-authentication capabilities could become strategic partners or acquisition targets as retailers build for AI-led checkout.
What to watch
- Visa, issuers or Indian payment partners launching production-grade agent identity, credentialing or delegated-payment products.
- RBI, NPCI or card-network guidance on liability, consent, recurring mandates, authentication and dispute handling for AI-initiated transactions.
- Major Indian marketplaces, quick-commerce platforms, travel sellers or large retailers enabling authenticated agent checkout.
- Merchant payment gateways adding agent-detection, tokenized delegated authorization or agent-specific fraud-management features.
- Evidence that AI shopping agents begin sending material referral traffic or completing purchases rather than only assisting discovery.
- Public fraud, unauthorized-purchase or chargeback incidents involving autonomous purchasing agents.
- Retail APIs and product feeds evolving from marketing catalogs toward real-time availability, delivery promise and return eligibility data.
- Add structured, API-accessible product, inventory, price, promotion, delivery and return-policy data so shopping agents can reliably evaluate offers.
- Assess payment-stack readiness for network tokenization, delegated authorization, step-up authentication, device binding and auditable consumer consent.
- Define agent-specific fraud controls, including transaction caps, merchant-category restrictions, anomaly detection, cancellation windows and clear dispute workflows.
- Treat agent checkout as a new acquisition surface: measure whether product feeds, fulfillment reliability and returns performance affect agent recommendation and conversion rates.
- Prioritize pilots in repeatable, low-dispute categories such as replenishment, grocery staples, personal care and subscription-like purchases before high-ticket categories.
- Review marketplace and payment-provider contracts for ownership of customer data, agent identity signals, liability allocation and access to transaction-level insights.