Agentic Commerce Explained for Australian Retailers

Agentic commerce is shopping where an AI agent searches for products, compares options and completes purchases for a person within limits that person sets. The customer decides what they need and how much freedom the agent has. For Australian retailers, this creates another way for customers to find your products and place an order.

Some shopping assistants already help people choose products. The ability to complete a purchase depends on the agent, the retailer and the payment connection.

What is Agentic Commerce?

Agentic commerce gives an AI agent permission to carry out shopping tasks for a customer. That permission can cover a single purchase or a recurring task under agreed rules.

Agentic AI is the wider category of AI that works towards a goal and takes actions using tools. Agentic commerce applies that behaviour to buying and selling. An agent that books appointments is agentic AI, while an agent that buys replacement supplies is doing agentic commerce.

The term machine customers describes software or devices that purchase on someone’s behalf. A machine customer still acts for a person or business that sets its authority. A recommendation alone doesn’t give it permission to spend money.

In a hypothetical example, a customer asks an agent to replace their usual coffee beans when supplies run low. The customer specifies the brand and quantity, then sets a spending limit. The agent can place a matching order, but a different blend needs permission if substitutions fall outside those rules.

How an Agentic Commerce Agent Works

An agentic commerce agent turns a customer’s goal into a product search, then checks the options against the customer’s requirements. It reads store information through product feeds, connected tools or web pages.

The customer might ask for a waterproof jacket in their size that can arrive before a trip. The agent needs usable size information and stock, plus delivery terms for the destination. It compares suitable products and explains the trade-offs, such as a lighter jacket with less insulation.

Where the shopping service supports purchases, the agent can complete payment using delegated payment credentials within the customer’s permission. The separate guide to agentic checkout explains the transaction itself.

Several protocols connect different parts of this process. They do different jobs and can work together:

• Universal Commerce Protocol (UCP) gives agents and commerce systems a shared way to handle shopping capabilities, including buying and order support.
• Agentic Commerce Protocol (ACP) connects merchants with AI shopping experiences, including product discovery and supported checkout integrations.
• Agent Payments Protocol (AP2) records the customer’s authorised intent so a proposed purchase and payment can be checked against that permission.
• Model Context Protocol (MCP) connects an AI application to external data and tools, such as a catalogue search service.

A protocol describes how systems communicate. It doesn’t make every store compatible or give an agent unrestricted authority. The UCP explainer covers its role in connecting shopping systems.

How Agentic Commerce Differs From Ecommerce

Traditional ecommerce puts the customer in charge of browsing and clicking through the store. Agentic commerce delegates some of those actions to software while the customer sets the goal and limits.

Part of the purchase	Traditional ecommerce	Agentic commerce
Browsing	The customer explores product pages with search and filters	The agent searches against the customer’s requirements
Product information	Page design and product copy help a person decide	Readable attributes, current stock and explicit terms help the agent compare
Choosing	The customer compares options themselves	The agent can explain a shortlist or choose within permitted rules
Checkout location	Usually the retailer’s site or app	The retailer’s checkout, an assistant or another supported shopping interface
Control	The customer approves actions through the interface	The customer defines which actions the agent can take

The store still supplies the goods and handles the order. A different shopping interface changes how the customer reaches the store, not whether the product must meet the customer’s requirements.

Shopping Agents and the Agentic Web

Shopping agents are part of the agentic web, where AI agents research information and carry out tasks across sites for a person. The same customer could use agents to research a trip, book accommodation and buy suitable luggage.

For a retailer, the agentic web adds two needs. Product information must be readable by agents, and the store must recognise agents that have permission to act. A product image can persuade a customer, but the agent also needs explicit details such as capacity, colour and availability.

Trusted-agent signals help distinguish an authorised shopping request from unwanted automation. Mastercard Agent Pay, for example, includes registration and verification of trusted agents. Mastercard’s explanation of Agent Pay describes that requirement.

A store needs both discovery and trust. An agent that reads the catalogue still needs the right permission before it can buy.

What is Driving the Rise of Agentic Commerce

Agentic commerce is developing because AI assistants can interpret shopping goals, while commerce platforms and payment networks are connecting them to real transactions. Product research is already a task people can perform in everyday assistants such as ChatGPT and Google’s AI Mode.

Platforms are opening catalogue information to those assistants. OpenAI describes ACP product feeds and Shopify Catalog integration as ways to supply products for ChatGPT discovery. Google uses its Shopping Graph to connect conversational searches with product information.

Payment networks are addressing consent and transaction security. Visa Intelligent Commerce uses tokenised credentials and customer-defined limits. Visa’s explanation of agentic shopping distinguishes browsing from permission to buy. Mastercard Agent Pay also defines consumer control and a process for disputes.

These connections reduce the need for every agent to build a separate shopping arrangement with every retailer. Adoption still depends on supported markets and compatible store systems.

Where You Can Use Agentic Commerce Today

Australian shoppers can use AI assistants for product research now, while agent-completed purchases depend on the shopping service and retailer. These are different levels of shopping assistance.

Google’s AI Mode is available in Australia and can draw on shopping data when answering questions. Google has also announced Australian expansion of UCP-powered checkout. Its newer integration-hub capabilities are rolling out in the US, with Australia to follow, so an announcement doesn’t mean every Australian store supports them.

Australian retailers can use Merchant Center’s AI performance insights to see how their brand appears across Google’s AI surfaces. Google’s retailer update describes the available markets and the separate UCP rollout. The guide to Google’s AI shopping features in Australia covers those channel differences.

ChatGPT helps shoppers browse products and compare options in a conversation. OpenAI has moved its focus from the initial Instant Checkout experience towards product discovery and merchants’ own checkout experiences. Shopify products can appear through Shopify Catalog, with purchases completed on the merchant’s store through an in-app browser. A product appearing in ChatGPT doesn’t mean ChatGPT automatically buys it.

Agentic AI in retail also works on a retailer’s own site. A store agent can help a customer find a product and refine their choice, then take supported basket or order actions. Retailers can also use agents internally to answer staff questions about their product range or assist with merchandising. These uses don’t require the customer to delegate purchases across the open web.

Benefits and Risks for Businesses

Agentic commerce can bring retailers demand from customers who start their shopping in an AI assistant. A product that matches a detailed request can reach a customer who hasn’t visited the store before.

Shopping assistance can also remove reasons to abandon a basket. An agent that answers a delivery question or finds an available size helps the customer continue. That benefit depends on accurate information and a working purchase flow. It isn’t a guaranteed increase in sales.

The risks follow from handing part of the purchase to software:

• Customers can spend less time with your site and brand when an external assistant handles the comparison.
• Stale stock or an incorrect product attribute can lead to an unsuitable recommendation or an order your store can’t fulfil.
• An agent can misunderstand a request, especially when a customer leaves a requirement unstated.
• A disputed purchase needs a clear record of what the customer authorised and what the agent ordered.

The location of the agent changes brand contact. An assistant on your site can keep the conversation with your store, while an external agent can compare you with other retailers. Both need accurate product information and a route to human help when the order needs it.

The Future of Agentic Commerce

Over the next two years, Australian retailers are likely to see more shopping journeys begin with a customer’s request to an AI assistant. That outlook follows the catalogue connections and payment standards being developed. Customers will still decide how much authority to give each agent.

One likely change is more demand for detailed product attributes. Agents comparing a jacket’s waterproofing or a sofa’s dimensions need those facts explicitly. Another is more attention to permission records when an agent places an order or changes a basket.

Discovery and payment won’t necessarily expand at the same rate. ChatGPT’s shift towards merchant checkout shows that product research can grow while the purchase stays on the retailer’s site. Google’s announced Australian checkout expansion creates another possible route, with availability tied to its rollout and merchant participation.

For your store, the practical decision is where you want customers to receive shopping help and which tasks you’re willing to delegate. Start with the guide to preparing a store for agentic commerce to turn that decision into catalogue and store changes.