Visa CMO: AI agents are your new customers — here’s how to sell to them

The article argues that AI agents are emerging as a new customer segment, with Visa research showing 71% of businesses willing to optimize products and offers for AI agents and more than half open to agent-to-agent price negotiation. It highlights a shift toward machine-readable product data, structured knowledge bases, live pricing/availability interfaces, and trust signals as key competitive requirements. The piece is strategic and forward-looking rather than company-specific, so near-term market impact is limited.

Analysis

This is directionally bullish for network-native payment and commerce rails because AI-mediated shopping increases the value of a system that can authenticate, price, settle, and resolve disputes with minimal human friction. The near-term monetization is not in consumer-facing AI search per se; it’s in the plumbing: tokenized credentials, merchant data normalization, fraud screening, and orchestration layers that sit between agents and checkout. Visa is the obvious public-market proxy, but the second-order benefit is broader: any issuer/acquirer stack that can reduce false declines and improve machine-readable merchant quality scores should win share as agents become a higher-intent traffic source.

The bigger second-order effect is a bifurcation of commerce winners into “legible” and “opaque” merchants. Brands with clean inventory, price, fulfillment, and policy data will gain disproportionate conversion because agents will optimize for certainty, not brand aspiration. That compresses the advantage of heavy spend on upper-funnel marketing and shifts budgets toward data hygiene, feed management, and API reliability; by contrast, retailers and marketplaces with messy catalogs, stale inventory, or inconsistent shipping promises risk traffic leakage even if demand is unchanged.

The market may be underestimating the cybersecurity angle. If agents can negotiate and execute, then identity, permissions, and payment credentials become a new attack surface; expect rising demand for transaction-level anomaly detection, consent management, and agent-authentication tooling over the next 12-24 months. A useful contrarian point: this trend is not uniformly bullish for all AI-search beneficiaries, because the value migrates from discovery to execution. Search aggregation may capture attention, but the durable rent likely accrues to infrastructure and the merchant-of-record layers that control trust, compliance, and settlement.

For Visa specifically, the upside is incremental rather than transformational in the next two quarters, but the strategic optionality is real if AI-driven commerce meaningfully increases card-not-present volume and lowers fraud through better authentication. The risk is that agent-led checkout gets disintermediated by closed ecosystems or alternative payment rails that bundle identity and execution more tightly than cards. Still, in any scenario where agents transact more often, the default winner is the incumbent network that can standardize trust at scale.

Key Decisions for Investors

• Long V on a 6-12 month horizon; add on pullbacks if management commentary confirms agentic commerce pilots or API/credentialization initiatives. Risk/reward is favorable because this is an option on higher payment frequency plus better authorization economics, not a near-term multiple expansion story.
• Pair long V / short a basket of lower-quality retail and marketplace names with poor catalog hygiene over 3-9 months. The thesis is that AI-mediated demand routes toward the most machine-readable merchants, widening conversion dispersion and exposing operational slippage.
• Buy cybersecurity infrastructure exposure on weakness over 6-18 months, favoring names tied to identity, fraud, and consent management rather than generic endpoint. The upside is a multi-year expansion of the threat surface as agents gain transaction authority.
• Avoid overpaying for pure AI search beneficiaries; use rallies to fade names whose monetization depends on discovery clicks rather than execution control. If the market is pricing all agentic traffic as incremental traffic, that is likely too optimistic.
• Monitor merchant-side software and payments processing names for a selloff on 'AI search disintermediation' headlines; selectively buy only those with strong data normalization and checkout integration, since they may become the toll collectors rather than the disrupted layer.