For years, real estate portals have competed to attract buyers through the size of their listings databases, the quality of their search tools and the amount of information they hold about consumers.
But a new analysis published in Harvard Business Review suggests the next competitive advantage may belong to neither the portal nor the business selling the property, but to the artificial intelligence assistant acting on behalf of the customer.
In the recent article, AI and platform strategy adviser Sangeet Paul Choudary argues that consumers are increasingly delegating the discovery and comparison of products to AI assistants, potentially weakening the influence of the marketplaces they have relied on for decades.
The shift is already becoming relevant to Australian property searches. REA Group launched a realestate.com.au app inside ChatGPT in February 2026, while Domain has introduced AI-powered interactive floorplans and expanded its virtual property viewing capabilities.
Although the approaches differ, both developments illustrate how technology is changing the way buyers find and assess properties.
For Australian real estate professionals, the question is not simply whether AI will become part of the property search, but who will influence the decisions buyers make before they contact an agent.
Sangeet, a senior fellow at the University of California, Berkeley, and author of Reshuffle, argues that businesses can continue attracting customers and facilitating transactions while losing influence over the decisions that bring those customers to them.
Property portals are already part of the shift
Sangeet points to developments in the United States, where major property platforms have begun making their listings accessible through AI applications.
In October 2025, Zillow became the first real estate app available inside ChatGPT. By March 2026, rival Realtor.com had also launched a ChatGPT app, allowing consumers to access competing property services through the same conversational interface.
The development illustrates how the starting point for a property search could change. Rather than opening a particular portal and applying filters for location, price and property type, a buyer could ask an AI assistant to identify homes matching a much more detailed set of requirements.
Those preferences might include commuting times, nearby schools, renovation potential, property features or the suitability of a home for different stages of family life.
An assistant connected to multiple property services could potentially compare options without the buyer needing to move between individual websites.
For portals, that presents a challenge. Having the largest audience or the most familiar brand may become less decisive if consumers increasingly rely on an independent AI assistant to determine which properties deserve their attention.
Sangeet argues that when a single AI agent can search across competing marketplaces, those marketplaces risk becoming information sources rather than the places where consumers begin their purchasing decisions.
Why more website traffic may not tell the whole story
One of the more interesting findings highlighted in the Harvard Business Review article concerns the behaviour of consumers arriving at retail websites through AI.
Citing Adobe data, Sangeet reports that traffic from AI sources to US retail websites increased 393 per cent year-on-year in the first quarter of 2026.
Separate Adobe analysis found that visitors arriving through generative AI sources spent 32 per cent longer on retail websites, viewed 10 per cent more pages and were 27 per cent less likely to leave immediately.
On the surface, those figures suggest AI could become a valuable source of highly engaged customers.
However, Sangeet argues that stronger traffic does not necessarily mean a marketplace has greater control over the customer relationship.
An AI assistant may already have helped the consumer narrow their options, compare alternatives and decide what they want before they reach the website.
For real estate agents, that could eventually mean dealing with buyers who have completed much more of their research before making an enquiry; it could also change the importance of traditional online marketing measures.
Page views, enquiry numbers and portal engagement would remain useful, but agents and technology providers may need to understand more about how buyers arrive at their decisions and what happens after an enquiry is made.
The Adobe figures relate to US retail websites, not Australian property portals, but they offer an early indication of how AI-assisted consumers may behave differently.
The value of understanding what happens after the enquiry
Sangeet argues that one of the most valuable sources of competitive advantage in an AI-driven marketplace will be information about the decisions consumers ultimately make.
A property portal can track searches, saved listings, page views and enquiries.
But those activities do not necessarily explain why a buyer inspected one property and rejected another, why an offer was made or why a transaction failed to proceed.
In his analysis, Sangeet identifies the importance of understanding what happens after discovery, including which enquiries lead to offers, which offers become contracts and which contracts result in completed transactions.
For Australian agents, that raises an interesting point about the information already sitting inside their businesses.
Agency CRMs can contain years of buyer conversations, inspection feedback, appraisal histories, negotiations and transaction outcomes.
While portals may have broader visibility of online search behaviour, agents can hold valuable information about the decisions made during the actual sales process.
If that information is recorded accurately and used appropriately, it could become increasingly useful as AI systems develop.
An agency that understands not only what buyers search for, but why they buy or walk away, may be better placed to identify opportunities and provide relevant advice.
That does not mean an AI assistant can automatically access or interpret every conversation inside a CRM. The usefulness of the information would depend on data quality, system connectivity and appropriate privacy controls.
Nevertheless, the analysis points to a potential shift in the value of the information agencies collect.
Could AI weaken the advantage of the biggest platforms?
Another argument in Sangeet’s analysis concerns the competitive protection traditionally enjoyed by large marketplaces.
Platforms have historically benefited from network effects: the more buyers they attract, the more sellers want to participate, and the more sellers they attract, the more valuable the platform becomes to buyers.
That dynamic has helped major online marketplaces establish strong positions in their industries.
Sangeet argues that AI could weaken some of those advantages by allowing smaller competitors to offer sophisticated matching services without first building an equally large network.
He cites examples from employment and corporate purchasing, where businesses are using AI to extract more useful information from available data and improve how customers are matched with products or services.
The potential relevance to real estate is considerable. An independent agency or smaller property platform may not be able to compete with a national portal on audience size, but it could potentially use AI to offer more personalised buyer matching or a better understanding of local market conditions.
For example, an agency with detailed knowledge of buyer requirements and its local property market might be able to identify suitable opportunities more effectively than a system relying mainly on broad search criteria.
Whether that translates into a meaningful competitive advantage will depend on the information available and how well the technology performs.
Still, it suggests that the size of a database may not always be as important as the ability to understand and use the information within it.
What happens when AI starts acting for the buyer?
Perhaps the biggest change identified in the Harvard Business Review analysis is the movement from AI helping consumers research purchases to AI acting on their behalf.
Sangeet describes emerging AI assistants capable of completing forms, making bookings, negotiating and carrying out purchases.
In his article, he reports that Meta launched Muse, an AI assistant designed to perform tasks for consumers, in September 2026.
He also describes a subsequent dispute in which Amazon blocked Muse from shopping on its platform, arguing that third-party agents should identify themselves and obtain permission before conducting transactions.
The example illustrates the commercial tensions that can arise when AI systems begin interacting directly with businesses.
For real estate, the prospect of AI acting on behalf of buyers raises a different set of questions.
Could an assistant eventually manage property alerts, arrange inspections, compare comparable sales and communicate with agents about a buyer’s requirements?
Could it help buyers evaluate properties against their financial circumstances and long-term plans?
Some of these functions are technically plausible, but property transactions involve legal, financial and personal decisions that require substantially more oversight than routine online purchases.
The more immediate opportunity may be for AI to handle the research and administrative work surrounding a transaction while buyers and agents remain responsible for the important decisions.
Even that would represent a significant change to the way agents interact with prospective purchasers.
What it means for Australian real estate agents
The analysis does not suggest that property portals are about to disappear or that AI will replace real estate agents.
Rather, it raises questions about where influence sits as buyers gain access to technology that can search across multiple services and interpret information on their behalf.
For agents, the practical implications could extend to how listings are presented, how buyer information is maintained and how enquiries are managed.
Accurate, detailed and accessible property information may become increasingly important if AI systems are involved in deciding which homes to recommend.
Agents may also need to consider whether their technology can connect with emerging AI services, and whether their databases contain the kind of useful, structured information that those services require.
There is also a broader question about the role of the agent; if buyers arrive with more research already completed, the value of an agent’s advice may increasingly rest on the information and judgement that cannot be obtained from a property listing alone.
That includes understanding local market conditions, explaining competing buyer interest, navigating negotiations and helping clients manage the complexities of a transaction.
Those responsibilities are not removed simply because the buyer has a more sophisticated search tool.
What could change is the point at which the agent becomes involved and the information the buyer expects them to provide.
Sangeet’s central argument is that businesses will increasingly need to earn the right to represent customers rather than assume that controlling a marketplace guarantees control of the relationship.