AI shopping agents now research, shortlist and buy on a customer's behalf, and they read structured product data, not design. With Australians spending $82.6 billion online, retailers whose feeds, schema and pricing are not machine-readable will quietly fall out of the consideration set.
The Take: The shopper about to compare your product might not be a person. AI agents now research, shortlist and check out on a buyer's behalf. They do not care how your site looks. They read structured data. If your product feed, schema and pricing are not machine-readable, the agent skips you and you never know the sale existed.
Zoom out: Australians spent $82.6 billion online in 2025, up 14% on the year, with 24% of all retail now happening online. That pool is where AI shopping agents are learning to operate. ChatGPT alone is fielding roughly 50 million shopping queries a day and driving a 42% lift in conversion for the retailers it sends. The buyer is quietly changing from a human scanning a page to a machine parsing a feed.
The precedent: The plumbing is being laid now. OpenAI and Stripe are pushing one checkout protocol while Google and Shopify push another. After OpenAI deprecated its Instant Checkout program in March 2026 with fewer than 30 merchants live, it pivoted to owning discovery and handing the customer back to the merchant to convert. Translation: the agent picks the shortlist, you still have to close. If you are not on the shortlist, closing never comes up.
Spent online by Australians in 2025, the pool AI shopping agents are now learning to shop. Source: Australia Post, 2026
The fine print: Being agent-ready is not a rebrand. It is boring back-end hygiene. Clean product titles, complete structured data, accurate stock and pricing a machine can trust. The Australian retailers who treat their feed as seriously as their homepage will get shortlisted. The ones relying on a clever hero image and a human to admire it will fall out of the consideration set without a single bad review.
For Australian operators: Audit your product feed the way an AI agent would read it. Go field by field and fix the gaps in titles, attributes and pricing first. Make sure your structured data is complete and current, because that is what the agent parses before it ever considers you. Keep stock and price accurate to the minute, since an agent that gets burned once stops trusting the source. Treat machine-readable product data as a sales channel now, not a technical chore for later.