Australians will let AI choose. They won't let it pay.

Australian shoppers have already worked out what they want AI to do for them, and what they don’t. They are happy to let an agent compare products, check prices and narrow the list. They are not handing over the credit card. The gap between those two things is much wider than most of the coverage suggests, and it points somewhere useful if you sell online. A lot of merchant effort right now is going into the half that customers are most reluctant about.

Where shoppers have drawn the line

62% of Australian consumers are open to an AI agent helping them make purchasing decisions, according to the commercetools Agentic Commerce Radar, across everyday categories like household items, subscriptions, gifts and travel bookings.

Then the numbers fall off a cliff. In the same research, 65% of consumers trust AI to compare prices for them. Only 14% trust it to place the order.

People want the research done. They want to keep the payment.

Merchants are building the other half

Checkout.com surveyed merchants and consumers across six markets. 42% of merchants are already testing agentic commerce. 72% agree consumers will adopt agent-led shopping faster than most merchants are ready for.

On the consumer side, 33% expect at least 10% of their purchases to be AI-driven within a year. So the demand is real and it is arriving quickly.

But a lot of that merchant investment is going into autonomous checkout, which is exactly the part shoppers are most reluctant about. One in four say they will never delegate a purchase to AI. 27% trust no organisation at all to run a shopping agent for them.

What people want before they hand over a card

Checkout.com also asked consumers what would make them comfortable delegating spend. The answers came back almost level: spending caps, the ability to revoke permission instantly, and easy cancellation.

No single feature wins, which is the finding. People want control, and they want it visible.

The move I’d make first

The part of this that is already happening is discovery. AI is comparing products right now and recommending some of them. That runs on whatever your site publishes: product data, pricing, stock, reviews, delivery terms.

If that information is thin, stale or contradicts itself, the agent skips you and recommends someone clearer. No checkout protocol required, and no new platform to buy.

So before anyone builds agentic checkout, get the basics machine readable:

• Product feeds that are accurate and actually current, not a quarterly export
• Structured data that matches what a human sees on the page, including price and availability
• Real stock and delivery information, not “contact us”
• Reviews and specifications a machine can parse rather than an image of a spec table
• Consistent product naming across your site, your feed and your marketplace listings

That is unglamorous work. It is also the work that pays off whether agentic checkout arrives next year or in five, because the same clean data feeds Google Shopping, your own search, and every comparison tool in between.

If your product data is already in decent shape, this may be a small job. We will tell you that rather than turn it into a project.