Chosen by the Machine: Why Winning AI's Recommendation Is Now a Trust Problem, Not a Tech One

The Shift Nobody Sent a Memo About

A new layer now sits between your brand and your customer. Call it the AI decision layer. Before a person compares options, reads a review or clicks anything, an AI assistant has already scanned the field, decided who's relevant and credible, and handed back a shortlist. If your brand isn't on it, you're not losing the sale. You never entered the room.

We've watched this happen faster in Australia than most boards expected. According to Roy Morgan, 13.6 million Australians aged 14 and over, about 58% of that group, used AI tools in an average four-week period in the first quarter of 2026. ChatGPT alone reached 10.5 million of them. That's not an early-adopter niche. That's most of your market, using AI the way they used Google five years ago.

Our take at Bushnote: the moment AI use crosses half the population, "being on page one" stops being the goal. The goal is being the answer the assistant gives when there is no page, no list of ten blue links, just a recommendation. That's a different game, and it rewards different things.

This Is Happening Now, Not in Some Future Keynote

The buying behaviour has already moved. According to PayPal's Australian research, 48% of Australians have used AI assistants for online shopping searches, and that rises to 66% among people under 45. Looking forward, 78% expect AI shopping tools to become a mainstream part of shopping, and 53% plan to use them within the year.

The payment rails are arriving too. In January 2026, Mastercard completed what it described as Australia's first authenticated agentic transactions: an AI agent used a Commonwealth Bank debit card to buy cinema tickets from Event Cinemas, and a Westpac card to book accommodation in Thredbo. Mastercard has since put dedicated agentic commerce staff in its Sydney tech hub. Some estimates it cites put agent-led commerce at up to A$670 billion in Australian spending by 2030.

So the infrastructure question is settling. Machines can now find products, compare them and pay for them. What isn't settled, and where the real story sits, is trust.

The Trust Gap Is the Opportunity, and It's Specific

Here's the number every Australian leader should sit with. According to a Logica Research study for Commerce and PayPal, 63% of Australian shoppers are interested in trying AI shopping tools, but only 4% want AI to help at the point of payment. Australians are happy to let AI research and recommend. They pull their hand back sharply at the checkout.

PayPal's data tells the same story from another angle: 61% of Australians say they'd trust AI to make product recommendations, yet 92% hold at least one concern about AI shopping, and 64% worry specifically about privacy and the security of their details.

Read those two findings together and the strategy writes itself. The battle right now isn't for the autonomous purchase, which most people won't hand over yet. The battle is to be the brand AI surfaces during research and comparison, because that's the stage Australians have already opened the door to.

And there's one figure that decides who wins it. According to PayPal, 30% of Australians say they'd only use AI shopping solutions from brands or platforms they already trust. That single line is where our work and this whole shift meet. When a customer will only accept AI's help involving a brand they already believe in, the brands that win the recommendation are the trusted ones. Not the ones with the cleverest technical setup. The trusted ones.

We'd argue that's true for brands as much as for payment networks. Trust is the product now.

This Is a Reputation Problem in Disguise 

There's a comforting misreading of all this going around: that AI visibility is a plumbing problem. Get the technical setup right, make the site machine-readable, and you'll be fine. Fix it and forget it.

We think that's half the picture, and the less important half.

Yes, a machine has to be able to read you. But once it can, the question it's answering is the same one a person asks: can I trust this brand? AI weighs the signals we've always associated with reputation. Does the brand say the same thing about itself everywhere? Does it show real expertise, or just marketing noise? Do the reviews, the listings, the pricing and the third-party mentions line up, or do they contradict each other? When those signals conflict, the machine's confidence drops, and a lower-confidence brand quietly slips off the shortlist.

That's not an IT job. Consistency of story across every channel, genuine authority, a reputation that holds up wherever it gets checked: this is brand and narrative work. It's the work marketing and communications teams have always owned. AI has just turned it into visibility infrastructure and raised the stakes.

Our position at Bushnote is blunt: if you treat being chosen by AI as a technical task, you'll optimise your way onto a list you then fall off, because the thing deciding your place is judging your credibility, not your code.

Picture two businesses in the same category. One has a spotless technical setup and a website that reads well, but its story wobbles: the homepage promises one thing, the reviews tell a mixed story, the pricing looks different depending on where you find it, and there's little independent evidence it's any good. The second has a plainer site, but everywhere the machine looks, the same clear claim holds up, backed by real reviews, consistent details and credible third-party mentions. Ask an AI to recommend a provider, and it's the second business that gets named. The first optimised for the machine's eyes. The second earned the machine's trust. That gap is the whole argument.

What the Machine Actually Checks

You don't need to understand how a large language model works to influence what it recommends. You need to understand what it's looking for, which is roughly what a careful person looks for.

One story, told the same way everywhere. AI reads your site, your listings, your reviews, your press and your social profiles as one picture. If your positioning is sharp on the homepage and mush everywhere else, the machine sees an inconsistent brand and hedges. Coherence is a ranking factor now, even though nobody calls it that.

Proof of real expertise. Generic content that could've come from anyone gets treated as filler. Original insight, genuine experience and a clear point of view are what get you quoted and recommended. This is where thin "content at volume" strategies fall down.

Signals that agree with each other. Your prices, your availability, your business details and your reputation across the web should tell one consistent story. Contradictions read as risk. Risk lowers confidence. Lower confidence means you're left off.

A reputation that survives a second look. When AI, or a cautious buyer, checks you against independent sources, what comes back needs to reinforce your claims. Trust built only on your own channels is fragile. Trust echoed by others is durable.

None of that is a schema tweak. All of it is strategy.

What Australian Leaders Should Actually Do

We'll keep this practical. Four moves, none of which require you to become technical.

Own one clear brand story, and make it consistent everywhere AI can read it. Decide what you stand for, say it in plain language, and repeat it across every surface. The machine rewards the brand it can describe in one confident sentence.

Build genuine authority, not volume. Publish things only your business could say, backed by real experience and evidence. One credible, original piece beats twenty generic ones for getting cited.

Align your signals across the web. Audit what the internet says about you: reviews, listings, pricing, third-party mentions. Fix the contradictions. You're not just tidying up; you're raising the confidence score a machine assigns you.

Treat "does AI recommend us?" as a board-level metric. Rankings and traffic still matter, but they no longer tell the whole story. Ask, regularly and directly: when a customer asks an AI in our category, do we come up, and what does it say about us? Track that answer over time the way you track anything else that decides revenue.

A word of caution on that last point. According to Adobe, e-commerce traffic from generative AI platforms in Australia jumped several hundred per cent over the last holiday period. As AI handles more of the discovery, your direct traffic can fall even while AI-influenced revenue climbs. If you're only watching sessions and clicks, you'll misread a win as a loss and cut exactly the investment you should be growing.

The Honest Counterpoint

We should be fair about what the sceptics get right. Australians are still cautious. The 4% who want AI at the checkout, and the 92% carrying at least one concern, are a real brake on how fast full agentic commerce arrives. Fully autonomous shopping isn't the default this year, and anyone selling it as already-here is overreaching.

But caution about the checkout doesn't slow the part that matters for you. Discovery and comparison have already moved. The recommendation is already being made. Waiting for autonomous checkout to mature before you act means sitting out the years when the trusted brands are quietly being chosen and the untrusted ones are quietly being filtered out. By the time the checkout question resolves, the shortlist habits will be set.

Where Bushnote Fits

We've spent our work shifting perception and shaping decisions: for brands under pressure, for organisations navigating public opinion, for causes that need to be believed as well as heard. Being chosen by AI is the same problem in a new venue. It rewards a coherent story, real authority and a reputation that holds up under scrutiny, applied consistently wherever the machine can look.

That's the work. Not gaming an algorithm, but becoming the brand a careful reader, human or machine, has good reason to trust. Move silently, make impact.