Agentic commerce: when your next customer is not a person

In March 2026, Visa launched a program called Visa Agentic Ready across Europe, with 21 banking partners including Barclays, HSBC, Revolut, and Banco Santander. The program gives banks a structured way to test payments initiated not by a person tapping a card or clicking a button, but by an AI agent acting on that person's behalf. In Latin America, Santander and Visa have already completed live pilot transactions where AI agents purchased goods across five countries.

This is not a concept paper or a roadmap slide. It is happening. And for any regional business that sells products or services online, or plans to, it is worth paying attention to what it means.

What agentic commerce actually is

The term "agentic commerce" describes a scenario where a consumer gives an AI assistant a task ("order more fencing wire when the price drops below $X" or "book the cheapest return flight to Melbourne next Tuesday") and the AI goes and does it. The agent searches, compares, selects, and completes the purchase. The human approves the transaction, but the shopping, the comparison, and the selection happen without them.

This is different from the AI chatbots that businesses have been dealing with for years. Those are customer-facing tools that help people find products or answer questions. Agentic commerce puts AI on the buyer's side of the transaction. The AI is the customer's representative, and it behaves differently from a human browsing a website.

An AI agent does not respond to brand loyalty, visual merchandising, or emotional appeals. It reads structured data. It compares prices, specifications, availability, and delivery terms systematically. It does not get distracted by a banner ad or browse related products. It executes a task.

Why this matters for regional businesses

If you sell agricultural inputs, equipment, produce, accommodation, professional services, or anything else through an online channel, the way your business appears to an AI agent is going to matter. Not tomorrow, but soon enough that it is worth thinking about now.

Consider a farm supplies business in the Wimmera that sells fencing materials, chemical, and hardware through an online store. Today, most of their online customers are people who browse the catalogue, maybe compare a few options, and place an order. In an agentic commerce world, some of those customers will send an AI agent to do the browsing for them. The agent will look for structured product data: exact specifications, current pricing, stock availability, delivery timeframes to a specific postcode. If that data is not machine-readable, the agent moves on to a supplier where it is.

The same logic applies to a bed and breakfast on the Bass Coast, a seafood wholesaler in Lakes Entrance, or a consulting firm in Ballarat. If the information about what you sell, what it costs, and how to buy it is buried in PDFs, locked behind phone calls, or only available through a form that requires human interaction, an AI agent cannot transact with you. You become invisible to a growing channel.

What is real and what is not

It is important to be clear about the timeline. Agentic commerce is not going to replace human shopping next month. The Visa pilots are controlled tests with specific banking partners. The infrastructure for authentication, consent, and dispute resolution is still being built. Visa has released an experimental command-line interface for developers to test machine-initiated payments, which gives a sense of how early the plumbing still is.

In Australia, there are no announced pilot programs yet, though Visa has indicated that Asia Pacific pilots are planned for 2026. The regulatory and consumer protection frameworks for AI-initiated transactions in Australia have not been established. Questions about liability (if an AI agent buys the wrong product, who is responsible?), consumer guarantees under Australian Consumer Law, and data privacy are all unresolved.

So this is not something to panic about. But it is something to prepare for, because the pattern of technology adoption is predictable: it starts with large retailers and enterprise buyers, then moves down the chain. By the time it reaches the point where a grazier in the Riverina can tell their AI assistant to reorder supplementary feed when prices hit a threshold, the businesses that are ready to transact with that agent will have an advantage over those that are not.

The practical implications

The good news is that most of what makes a business ready for agentic commerce is the same thing that makes it ready for good e-commerce in general. Structured product data with clear specifications. Accurate, real-time pricing and stock information. Clear delivery terms. A checkout process that does not require a phone call or a custom quote for standard items.

For regional businesses, the specific gaps tend to be around data structure. Many smaller operations have websites that function more like brochures than transactional platforms. Product information lives in the owner's head, in a spreadsheet, or in a catalogue PDF. Pricing might be "call for a quote" rather than listed. These are not just barriers to AI agents. They are barriers to any form of online commerce. But agentic commerce raises the stakes because an AI agent will not call for a quote. It will simply go elsewhere.

There is also a less obvious implication around how businesses describe what they sell. AI agents rely on structured data, but they also use the plain-text descriptions on product pages to understand what something is and whether it matches the buyer's request. Writing clear, specific, jargon-free product descriptions matters more when the reader might be software rather than a person. A description that says "premium quality, industry-leading solution" tells an AI agent nothing. A description that says "6mm galvanised steel cattle panel, 2.1m x 2.4m, suits standard portable yard systems" tells it everything it needs.

What to do now

Nobody needs to overhaul their business this week. But there are sensible steps that are worth taking because they improve your online presence regardless of whether agentic commerce takes off quickly or slowly.

Look at your product or service listings through the eyes of a machine. Is the information structured, specific, and current? Are prices listed or hidden behind a quote request? Can someone (or something) complete a purchase without picking up the phone? If you use an e-commerce platform, check whether it supports structured data formats like Schema.org markup, which help both search engines and AI agents understand what you sell.

Generative AI tools can help with this work directly. Claude or ChatGPT can review your product descriptions and suggest improvements for clarity and specificity. They can help generate structured data markup for your website. They can draft the kind of clear, specification-focused product descriptions that both human customers and AI agents find useful.

The shift toward agentic commerce is a continuation of a trend that has been building for years: the move from relationship-based selling to data-driven discovery. For regional businesses, where relationships and local knowledge have always been strengths, that can feel threatening. But the businesses that will do best are those that combine their local expertise and relationships with the kind of clear, structured, accessible information that the next generation of buyers (human and otherwise) will expect.

If you want to talk through what this means for your specific business, get in touch. This is one of the shifts we are watching closely at ARAIN, and we are happy to help regional businesses think it through.