Google is turning search into a place where an AI agent finds, compares and buys on a shopper's behalf. The businesses whose products the agent can actually read will win the sale. The rest will simply be skipped.
Listen to this article
0:00 / --:--Narrated by Margot Ellis
Search is quietly changing from a place people visit into a place that acts. Google has been building the plumbing for it in the open: its AI Commerce Search documentation describes a system where an AI reads a store's structured product data, its prices, stock, attributes and fulfilment options, and uses that to search, compare and recommend on a shopper's behalf. Search Engine Land reported on 10 July 2026 that this is coalescing into a single open standard, a Universal Commerce Protocol, letting AI agents discover and complete purchases entirely inside AI experiences like Gemini and AI Mode.
Strip away the jargon and it comes to this. The old path was query, click, buy: a customer typed something, landed on your website, had a look around, and decided. Every step of that was a chance for you to win them, your photos, your reviews, your story, your shopfront. The new path is different. The customer asks their assistant to find and order the thing, and the assistant does the searching, the comparing and the buying itself. The person often never visits your site at all. They just see the shortlist the AI handed back.
For an Australian small business, that is either the best shelf you have ever been on or one you are quietly missing. It depends entirely on whether the agent can read you.
Think about how an AI agent actually shops. Someone says, find me a replacement filter that fits my fridge, in stock, delivered fastest. The agent does not browse the way a person does. It cannot be won over by a nice hero image or a heartfelt about page. It works from structured facts: what you sell, the exact model it fits, the live price, whether it is genuinely in stock right now, and how quickly you can get it there. If those facts are clean and machine readable, you are a candidate for the sale. If they are buried in a PDF, out of date, or simply not published in a form the agent trusts, you are not skipped out of malice. You are skipped because you were unreadable.
This is why the shift matters more than another search update. Duane Forrester, who built Bing's webmaster tools and now writes about generative search, has been making the point for months: you can no longer see how the AI ranks you, so the only lever you control is being something it can cleanly read and trust. When a machine is the one comparing options, legibility beats persuasion.
It is tempting to file this under IT and hope your website platform sorts it out. It will not, at least not on its own. Being present in agentic commerce is the same discipline as getting cited in AI answers, just with money attached to the outcome. It is about making sure the facts a machine needs, your products, prices, availability and terms, are published clearly, kept accurate to the minute, and structured the way these systems expect. That is unglamorous, ongoing work, and it is precisely the sort of thing that falls through the cracks when you are busy actually running the shop.
The upside is that the field is wide open. Most small businesses have not thought about this at all yet, which means the ones that get ready early get recommended while their competitors are still invisible. This is the same pattern we saw with ordinary search fifteen years ago and with AI answers over the past two years: the businesses that move before the crowd own the position, and the ones who wait pay to claw it back later, if they can at all.
You do not need to understand the protocol to benefit from it, any more than you needed to understand HTTPS to sell online. What you need is for your store to be legible and trustworthy to the agents doing the shopping. Handled properly, that is what good looks like, and it sits right alongside the agents that can already book and buy on your own site:
When an AI does the shopping, it cannot fall in love with your brand. It can only recommend what it can read. Being legible is the new being findable.
None of this means the human customer disappears. It means a new gatekeeper has appeared in front of them, and that gatekeeper judges you on clarity, accuracy and trust rather than charm. The businesses that treat their product data as a shopfront the machines walk past, and keep it immaculate, will get the recommendation. The ones who leave it messy will watch orders drift away without ever seeing the customer who chose someone else.
This is exactly the work we do at NextAura. We get Australian small businesses ready to be found and bought in the new era of AI search and commerce, making sure your store speaks the language these agents read so you are on the shortlist when it counts. If you would rather have this handled by people who track these shifts daily, get in touch and we will take it from here while you get back to your customers.
Tell us where you are headed. We will come back with a scope, a price, and a launch date you can plan around.
Book a free consultation
AI SEOOpenAI's new GPT-Live turns ChatGPT voice into a real conversation that searches the web mid-sentence. When the answer is spoken aloud, it names one or two businesses, not ten. Here is what that shift means for getting found.
AI SEOA new study of 6.77 million AI-driven visits found that nearly all the traffic AI now sends to websites comes through one door: ChatGPT. Here is what that means for where your next customers come from.
AI SEOGoogle's own data shows the average AI search is now triple the length of an old keyword query, and one in six uses voice or images. If your website still answers the old two-word search, it is answering a question nobody types anymore.