What Australia's UCP Checkout Rollout Means for Brands Selling Through AI Agents
By Malin Gaertig · September 11, 2026
Google chose Australia as its first Asia-Pacific market for Universal Commerce Protocol checkout, with Adore Beauty, Bunnings, Kogan Group, Petbarn and THE ICONIC already live. Read alongside agentic media buying, Singapore's governance-first rollout and McKinsey's build-vs-buy data, the week sketches where agentic commerce is actually heading — and what brands should do in the next two quarters.
- —Australia is Google's first APAC UCP checkout market — your product feed's machine-readability is now a competitive axis, not a technical detail.
- —Agentic media buying already runs without human checkpoints and stays inside guardrails — ask your agency who is accountable when an agent makes a bad call.
- —Singapore's governance-first, regulator-led model is the likelier Asia playbook — auditability will clear procurement faster than conversion lift.
- —A third of enterprises skipped at least one software purchase thanks to agentic coding — enterprise software pitches now compete with build cost, not just rival vendors.
- —Eighty per cent of purchases still happen in physical stores — build agentic readiness as a parallel track, not a replacement for the trust doing most of the work.
Australia doesn't usually get picked to go first. This week it did, and the implications reach well past retail into how every function from media buying to software procurement gets run over the next year.
Google chose Australia as its first Asia-Pacific market for Universal Commerce Protocol checkout, with Adore Beauty, Bunnings, Kogan Group, Petbarn and THE ICONIC among the named retailers now live. New Salesforce data shows agentic search as the first step in the shopping journey has grown two hundred per cent year on year here, and ninety-five per cent of enterprise retailers say they're familiar with agentic commerce — familiarity being the easy part. That's one of five stories this week that, read together, sketch out where agentic commerce is actually heading: not a single global rollout, but several regional ones, moving at different speeds, tested against a retail reality that hasn't caught up to the narrative yet. If you're setting budget or strategy for the next two quarters, this is the landscape you're setting it against.
Why does Australia going first on agentic checkout matter to global brands?
Australia going first changes who the reference case is. For the past decade, most retail technology shifts arrived here twelve to eighteen months after the US. This one didn't — Google picked Australia as its APAC beachhead, and the five retailers now live with UCP-powered checkout (a genuinely mixed cohort spanning beauty, hardware, marketplace, pets and fashion) are the ones the rest of the region will be benchmarked against.
The source: Salesforce's State of Commerce ANZ research and coverage from Inside Retail Australia, both published this month.
So what for brands: this is a discoverability problem before it's a conversion problem. If your product feed isn't machine-readable — clean titles, live stock levels, structured attributes an agent can parse without guessing — you're not losing on price or creative, you're not being seen at all. That's a genuinely new competitive axis, and it rewards the brands moving on it now, not the ones waiting for a wider rollout to feel safe.
What does it mean when AI agents start buying the media, not just the product?
Adweek reported this week on a campaign for wellness brand Rouge Care, where two AI agents — one planning, one buying — ran a connected TV campaign through PubMatic's AgenticOS with no human hand on the buying decision. Twenty-five thousand dollars in spend returned over one hundred and twenty-five thousand in attributable sales, roughly five times the money back. Separately, this year's Digiday Technology Awards finalists, announced the same week, read like a checklist of AI-ready infrastructure and first-party data handling — capabilities that used to differentiate a platform now simply qualify it for consideration.
So what for brands: the interesting part of the Rouge Care case isn't the return, it's that the decision loop ran without a human checkpoint and still stayed inside the brand's guardrails. That's the part that scales. Worth noting the IAB is only just starting work on a measurement framework for agentic advertising — the tooling is ahead of the standards, which is exactly the kind of gap that gets expensive once volume picks up. Get your own media agency answering two questions now: which parts of our buying already touch an agentic layer, and who's accountable when an agent makes a bad call unsupervised.
Is Asia's agentic infrastructure actually ahead of the West?
Two things landed in the same fortnight. AGNTCon and MCPCon Japan ran in Tokyo on 10 and 11 September, pulling in Anthropic, Google, Microsoft, SoftBank, Hitachi and NTT to work through how agent systems connect safely across the region. And Singapore's Monetary Authority committed two hundred and twenty million Singapore dollars over three years to its FinTech innovation scheme, with AI adoption and infrastructure as named tracks — building on the SAFR safeguards framework MAS published earlier this year governing how financial agents authenticate and act.
So what for brands: Singapore is positioning itself as the region's agentic commerce settlement layer, not just an adopter — real-time payment rails, a public safety framework, and funded runway behind it. The likely path for the region, most plausibly, is regulator-led infrastructure first and consumer-facing agentic shopping second, a different sequencing to the US pattern of shipping the product before the governance catches up. Any brand with an Asia-facing roadmap should stress-test assumptions about rollout speed against that governance-first posture — vendors who can demonstrate auditability will clear procurement faster than ones who can only demonstrate conversion lift.
Why are enterprises suddenly building software instead of buying it?
McKinsey's State of AI 2026 survey, widely covered this week, found close to a third of organisations decided against buying at least one software product because agentic coding tools let them build the functionality internally — rising to nearly half among the highest-performing companies, concentrated in technology and healthcare.
So what for brands: this is the quiet flip side of everything else happening in agentic commerce. While brands debate letting agents shop on their behalf, their own technology teams are using agents to quietly disintermediate the software vendors those brands rely on. Anyone selling software into the enterprise is now plausibly pitching against a product manager and a coding agent, not just the next vendor on the shortlist — and the pitch needs to justify itself against build cost, not just against competitors.
Are eighty per cent of purchases still happening in a physical store?
Yes — and that number, from this week's Mumbrella REmade coverage, is the useful corrective to the rest of this list. Agentic AI will most plausibly reward brands that already have trust, evidence and customer connection built up, and punish mediocre ones, rather than rewriting retail from scratch overnight. Everything above is real: live checkout infrastructure, funded regional rollouts, five-times returns on agentic media. None of it changes the fact that most revenue today still runs through a person, in a store, trusting a brand they already know.