Published on offmarket.now | April 2024
Off-market property has always been about who you know. A buyer's agent with the right connections, a whisper at an open home, a tip from a mortgage broker. For decades, finding properties that never hit Domain or realestate.com.au meant working the phones.
That is changing fast.
In the last twelve months, a new class of property tools has emerged that applies machine learning and large-scale data aggregation to the off-market problem. Instead of relying on personal networks, these platforms scrape hundreds of agency websites daily, geocode listings, and surface them to buyers within hours.
The mechanics vary by platform, but the AI layer on top is what makes these tools useful. Natural language processing cleans messy address formats. Geocoding APIs turn "3/42 Smith St Surry Hills" into a precise latitude and longitude. And postcode-based alert systems notify subscribed buyers the moment something new appears in their target area.
This is not theoretical. A growing number of platforms — including offmarket.now — are already surfacing off-market and pre-portal listings across Sydney.
The biggest signal that AI is reshaping property search came in late 2025, when REA Group rolled out natural language search to all realestate.com.au users after a 12-month pilot. Instead of clicking through dropdown filters for bedrooms, bathrooms, and price range, buyers can now type queries like "three-bedroom house with a pool in Marrickville under $2 million."
In February 2026, REA took it further by launching Australia's first ChatGPT property search app — a dedicated integration that lets users search for homes conversationally within the ChatGPT platform itself.
These moves signal a fundamental shift in how Australians discover property. The filter-based search paradigm that has dominated since the early 2000s is giving way to intent-based discovery. And for off-market properties — which often don't fit neatly into standard filter categories — this is significant. A buyer who types "renovator's delight near good schools in the Inner West" is expressing intent that rigid filters can't capture.
Perhaps the most consequential AI development for off-market is predictive seller identification. Tools like PropStream (which acquired BatchLeads in 2025) and SmartZip now use machine learning to analyse homeowner data — equity positions, length of ownership, life events, local market conditions — and predict which owners are most likely to sell in the next 6-12 months.
Fello's platform, which gained significant traction heading into 2026, takes this further by tapping MLS history to rank every property in a market by likelihood of sale. The implications for off-market are direct: if you can predict who's about to sell before they list, you can approach them first.
In Australia, REA Group has been developing its own "predicted serious buyer" metric, using AI to identify which buyer enquiries are most likely to convert. The supply-side equivalent — predicting which owners will become sellers — is the logical next step.
One of the persistent challenges in off-market transactions is valuation. Without the marketing campaign and competitive bidding that on-market sales generate, how do you know what a property is worth?
CoreLogic (now rebranded as Cotality since March 2025) has been pushing the frontier here. Their automated valuation models now achieve nearly 90% accuracy within 15% of actual sale prices across Australian markets. But the more interesting development is the emergence of multimodal valuation models.
A 2025 study applied a hybrid AI-BIM (Building Information Modeling) framework to a Melbourne high-rise, combining architectural data, natural language processing, and computer vision to value individual apartments. It achieved 100% accuracy within the range of recent comparable sales. The approach works by ingesting photos, floor plans, text descriptions, and spatial data simultaneously — exactly the kind of fragmentary information you get in an off-market deal.
The practical upshot is that the information asymmetry that has historically defined off-market property is eroding. Buyers no longer need to rely exclusively on agent relationships. They can:
None of this eliminates the value of a good buyer's agent. But it democratises access to information that was previously gated behind personal networks.
The convergence of these technologies — data aggregation, natural language search, predictive analytics, and AI-powered valuation — is creating a new category of property discovery tools purpose-built for the off-market segment.
Australia's off-market rate (the percentage of sales that occur without a public listing campaign) sits between 15-20% in major cities. That represents billions of dollars in transactions that most buyers never see. The platforms that can surface these deals efficiently, verify their value, and deliver them to the right buyers at the right time will capture an outsized share of attention.
The phone calls and whispered tips aren't going away. But they're being augmented — and in some cases replaced — by systems that never sleep and never forget a postcode.
---
Browse off-market listings or explore by suburb on offmarket.now.
See live off-market listings on the map
Browse listings →