Key Takeaways
- Visa's B2AI survey shows 53% of US businesses would allow AI agents to negotiate prices with other agents, and 88% are willing to share inventory and pricing data
- Consumers are comfortable with price comparison (58%) but reluctant about autonomous purchases (38%), revealing a clear gap with business sentiment
- Trust in bank-backed and payment network AI is roughly 1.3x higher than independent agents -- trust design will shape e-commerce strategy
Visa's B2AI Report Exposes the Business-Consumer Divide
AI-to-AI commerce looks poised to scale, with more than half of US businesses willing to allow agents to negotiate prices and terms directly with other agents on their behalf, according to a survey from Visa.
www.finextra.comOn April 2, 2026, Visa published its "Business-to-AI" (B2AI) report, conducted jointly with Morning Consult. The survey, fielded from January 29 to February 6, 2026, covered 2,000 US consumers and 512 business decision-makers, probing readiness for a world where AI agents participate in commerce as "customers."
The takeaway is straightforward: businesses are enthusiastic, consumers are cautious. The data makes clear that this perception gap is the primary bottleneck for agentic commerce adoption.
Why a Majority of Businesses Accept AI-to-AI Negotiation
The headline figure is that 53% of US businesses would allow AI agents to negotiate prices and terms directly with other AI agents on their behalf. Beyond that, 88% are willing to share pricing and inventory data with enterprise AI systems, and 71% say they would optimize products, offers, and experiences specifically for AI agents.
The fact that 77% are already using or piloting AI in their operations underpins this bullish stance. For these companies, transacting with AI agents is not some distant prospect -- it sits on the natural continuum of their existing AI adoption. That 55% already recognize the "B2AI" concept further illustrates how far along business preparedness has come.
The two payments giants are competitors, but Visa and Mastercard have common allies in their agentic commerce aspirations.
www.digitalcommerce360.comJack Forestell, Visa's chief product and strategy officer, told Digital Commerce 360 that he "hasn't seen anything like this since the dawn of ecommerce itself in the late '90s or early 2000s." With McKinsey & Company projecting $1 trillion in US agent-driven transactions by 2030, the urgency on the payments infrastructure side is well-founded.
The Invisible Line Consumers Draw
Consumer responses, meanwhile, follow a clear gradient. 58% feel comfortable letting AI compare prices, and 55% accept AI-applied discounts. But when it comes to "letting AI complete a purchase," the figure drops to 38%, and 60% say they would not allow an AI agent to spend any amount without prior approval.
This stepwise erosion of trust reveals that consumers draw a sharp distinction between "information gathering" and "spending decisions." They welcome comparison and recommendation but will not hand over the wallet. Visa CMO Frank Cooper III noted that "commerce is moving from market-to-human to market-to-machine," yet consumers are still in transit.
The generational split is striking. 48% of Gen Z trust payment network-enabled AI, compared to just 20% of Baby Boomers. Among Gen Z and Millennial AI shopping assistant users, roughly half report purchasing items they would not otherwise have considered. Across the full sample, about 40% of Americans say they have made an unplanned purchase through an AI tool -- evidence that agents are already reshaping buying behavior.
Trust Hinges on Who Operates the Agent
The survey surfaces a clear hierarchy of consumer trust. Bank-operated AI systems score 36%, payment network AI (Visa, Mastercard) scores 35%, and independent AI agents trail at 28%.
That 7-8 point gap signals that "who vouches for the agent" will be decisive for agentic commerce adoption. Visa's push for the Trusted Agent Protocol (TAP) under its Intelligent Commerce initiative, and Mastercard's counter with Agent Pay, are both bids to own this trust infrastructure. Visa is already working with over 100 partners globally, with 30+ building in the VIC sandbox, and has announced collaborations with Akamai, AWS, and Stripe.
Impact on E-Commerce Businesses
The survey findings carry three practical implications for e-commerce operators.
First, preparing agent-ready data is urgent. With 88% of businesses willing to share pricing and inventory data, any merchant that fails to structure its data risks being bypassed. Structured product data and real-time inventory and pricing APIs are now table stakes for being "discoverable" by agents.
Second, consumer-facing flows require a phased trust-building design. Letting agents handle price comparison and discount application while reserving final purchase approval for the consumer is the optimal hybrid approach for now. With 60% rejecting unapproved spending, deploying agent-led checkout without a human override is premature.
Third, prepare for both Visa and Mastercard agent payment standards. Both networks build on tokenization and cryptographic authentication, and betting on only one carries real risk. Starting tokenization infrastructure work now, incrementally, is the practical priority.
Summary
The picture drawn by Visa's B2AI report is clear. Businesses are already actively considering agent-to-agent transactions, while consumers remain in a "conditionally willing" posture. Bridging this gap is not a technology problem but a trust design problem, and the higher trust scores for bank and payment network AI confirm that "who guarantees the agent's identity" will govern the pace of adoption.
The key milestones to watch are whether Visa can deliver millions of agent-powered transactions by the 2026 holiday season, and how the standards competition with Mastercard resolves. E-commerce businesses need to respect consumers' invisible boundary while simultaneously building agent-ready data infrastructure and phased trust mechanisms.
