Australian real estate runs on a duopoly. realestate.com.au and Domain are where buyers browse, where campaigns are budgeted, and where a meaningful slice of every agency's marketing spend ultimately lands.
If you want your listing seen, you pay the portals: that has been the settled order for two decades, and nothing in this article suggests it's about to collapse.
But listings are only half the market. The other half, the half that decides which agent gets the campaign in the first place, is quietly moving somewhere the portals don't control.
When a vendor asks an AI assistant who should sell their home, the answer is not a portal search results page but a synthesis drawn from the open web: agency sites, reviews, local mentions, structured facts.
And on that layer, there is no duopoly. There's barely anyone at all.
What the portals actually own
It's worth being precise, because the opening only makes sense against the incumbency. The portals own listing distribution: the audience of active buyers, the campaign products, the alerts and the browsing habit.
That asset is real and durable: an AI assistant recommending an agent doesn't reduce anyone's need to advertise the property itself. What the portals have never fully owned is agent selection.
Vendors have always assembled their shortlists from a messier mix of reputation, word of mouth and research, and that messy mix is exactly what AI engines are now consolidating into direct, named recommendations.
Discovery is drifting off-portal
The behavioural shift is measurable. Compound questions (who's strong in this suburb, what's fair commission, auction or private treaty for a home like mine) used to fragment across dozens of searches and browsing sessions.
Now they collapse into one conversation with an assistant, which answers with names and reasons. The 2026 State of AI SEO in Real Estate industry benchmarks put a startling number on how mainstream that behaviour has become.
67%
of buyers now use AI as their primary research method before contacting an agent, per the 2026 State of AI SEO in Real Estate industry benchmarks
Overseas, the platforms can see the same drift: industry press in the United States has reported a partnership between Zillow and OpenAI to surface property search inside ChatGPT. Whatever shape such integrations eventually take in Australia, the direction is unambiguous: the conversation layer is becoming a front door to property, and the incumbents are taking it seriously.
Why this layer favours agents, for now
On a portal, an agent's visibility is substantially a function of products purchased and the platform's own logic. In AI answers, visibility is a function of the public evidence trail: explicit facts on your site, consistent identity across profiles, reviews, third-party mentions.
That's a game an individual agent or a small agency can win on merit and effort, without outspending anyone. It's also a game with a first-mover premium: the benchmarks indicate 91% of agents are effectively invisible in AI answers in their own market, which means most suburbs currently have no entrenched answer to displace.
Openings like that don't stay open.
How to claim the opening
The work is unglamorous and compounding. Keep running your portal campaigns: that's listing distribution, and it still matters.
In parallel, build the off-portal evidence trail: a website that states your suburbs, specialisation and track record in plain facts; identical identity details across every profile you hold; systematic reviews after every settlement; suburb commentary written in your own words. Then track the answers monthly, in your own suburbs, and tie the tracking to appraisals booked.
The wider context on how engines assemble these recommendations is in how AI search is changing agent selection, and the full agent programme is on our real estate industry page.
The strategic read
The honest framing is a new discovery layer forming above the existing distribution layer, not "portals versus AI" as combatants. Agents can't own the portals and never could.
The answer layer is different: for a limited window, it can be earned by whoever builds the best public evidence of being good at their patch.
The real risk is staying invisible while a competitor becomes the default answer in your suburbs, not backing the wrong platform.
No: they solve different problems. Portal spend distributes listings to active buyers; AI visibility wins the vendor decision that creates the listing. Most of the AI-visibility work is effort and consistency rather than media spend, so it rarely needs to cannibalise campaign budgets.
For listing-browsing questions, they already feature. But agent-choice answers are synthesised from the open web (reviews, agency sites, mentions), not from any single platform. Even if portal data feeds future integrations, your own public evidence trail remains the raw material engines reason over.
If anything, the opening is larger in regional markets, where fewer agents produce any citable content at all. An engine answering a vendor question in a regional town often has thin evidence to work with, which means a modest, consistent effort can dominate the answer quickly.
Unknowable precisely, but the mechanism is clear: once an engine has a well-evidenced default answer in a suburb, displacing it takes far more effort than claiming an empty position did. The rational move is to be early rather than to time it perfectly.