top of page

Only 23% Trust GenAI to Make the Payment. That Is the Agentic Commerce Problem.

Writer: Heather Fricke
Heather Fricke
Sep 10
2 min read

People are already comfortable letting AI help them find what to buy. Letting the machine actually spend the money is where everybody suddenly remembers that trust is not a decorative feature.

VISA JUST PUT A NUMBER ON THE TRUST GAP

On September 9, 2026, Visa released U.S. consumer research on agentic commerce. Visa reports that 72% of U.S. consumers have used an AI assistant, while only 23% trust generative AI to handle payment transactions on their behalf. When Visa is securing the transaction, reported trust rises to 61%.

That is not a small adoption detail. It is the commercial line between AI that suggests and AI that acts.

THE AGENT NEEDS MORE THAN INTELLIGENCE

An AI agent participating in commerce needs to know more than what product matches a request. It needs a reliable identity for the human or business it represents, explicit permission boundaries, a current understanding of the task, trusted evidence, and a record of what it was authorized to do.

Otherwise the system can be technically capable and commercially unusable. Nobody cares that the agent is brilliant if the buyer, merchant, payment network or business cannot answer the boring but expensive questions: Who told it to do this? Which identity is it acting for? What amount was authorized? What evidence did it rely on? Where does human approval begin again?

THIS IS WHERE MACHINE-READABLE AUTHORITY™ STOPS BEING A BRANDING CONVERSATION

Machine-Readable Authority™ is about making identity, evidence and authority legible enough for systems to distinguish what is true, current and permitted. In a search environment, that affects whether a business is understood and recommended. In agentic commerce, the same problem moves closer to the transaction.

Promptology™ matters here because an agent does not act on raw reality. It acts on the context it was given, the context it retrieved, the hierarchy it inferred and the decision boundary encoded around the task. That thinking environment is part of the commercial infrastructure now.

THE MONEY MOVES WHEN TRUST MOVES

The next wave of AI commerce will not be won by whoever can make the flashiest agent demo. It will be won by systems buyers trust enough to authorize. That means verified identity, explicit authority, permission controls, reliable context and human escalation are not compliance garnish. They are conversion infrastructure.

CHECK WHAT AI THINKS IS TRUE ABOUT YOUR BUSINESS BEFORE IT STARTS ACTING ON IT

Run the free AI Identity Check™ at https://www.frickeenergy.com/ai-identity-check. Frick-E Identity Engine™ is at https://www.frickeenergy.com/services.

PRIMARY SOURCE

Visa, “New Visa Research Finds Consumer Trust is Accelerating the Path to Agentic Commerce,” September 9, 2026: https://investor.visa.com/news/news-details/2026/New-Visa-Research-Finds-Consumer-Trust-is-Accelerating-the-Path-to-Agentic-Commerce/default.aspx

 
 
 

Recent Posts

See All
Published: AI Cited You. Did It Actually Use You?

The AI Journal published Heather Fricke’s contributed article, “AI Cited You. Did It Actually Use You?” on September 25, 2026. Why it matters The article separates citation presence from actual answer

 
 
 

Comments


bottom of page