I sat through a product demo last week from a PropTech company claiming their AI could match buyers to properties with 94% accuracy. Their algorithm analyses search behaviour, financial pre-approval data, lifestyle preferences, and historical purchase patterns to generate a ranked list of likely buyers for any given listing.

It was genuinely impressive. And I still don’t think it replaces what a good agent does.

Let me explain why — and where I think the real opportunity lies for agents who understand both the technology and its limits.

What AI Buyer Matching Does Well

Credit where it’s due: the data processing side of buyer matching has improved enormously over the past eighteen months.

Modern algorithms can ingest signals that no human could track at scale. How long a buyer lingers on a listing page. Which features they zoom into on photos. Whether they’ve viewed comparable properties in adjacent suburbs. Their mortgage pre-approval amount relative to their search price range. Whether they’ve attended open homes for similar properties.

Platforms like realestate.com.au and Domain already use versions of this internally to power their recommendation engines. The newer standalone tools go further by combining portal data with CRM records and social signals.

For high-volume agencies listing twenty or thirty properties simultaneously, this kind of automated matching is valuable. It surfaces potential buyers who might have been missed by a manual database search. It identifies cross-suburb interest that agents might not have noticed. It can flag buyers whose search patterns suggest they’re about to make a decision.

These are real, measurable improvements in efficiency.

Where AI Falls Short

Here’s the thing: buying a home isn’t a data problem. Not entirely.

The algorithm can tell you that a buyer has been searching for three-bedroom houses in Marrickville between $1.4M and $1.7M for the past eight weeks. It can predict with reasonable accuracy that they’re likely to be interested in your new Marrickville listing.

What it can’t tell you is that the buyer’s mother lives in Dulwich Hill and they really want to be within walking distance of her place. Or that they had a terrible experience with a previous terrace house and now they’ll only consider freestanding homes. Or that they’re actually waiting for one specific street to have a listing because the husband grew up there.

I deal with these kinds of nuances every week. The emotional, biographical, sometimes irrational factors that drive property decisions don’t show up in search data. They come out in conversations. At open homes. During those fifteen minutes when a buyer drops their guard and tells you what they actually want versus what they’ve been searching for online.

An experienced agent with strong relationships in their market carries contextual intelligence that no algorithm currently captures.

The Real Opportunity

I don’t see this as AI versus agents. I see it as AI doing the first pass and agents adding the layer that matters.

I’ve been talking to an AI consultancy focused on real estate applications about this exact question, and where I’ve landed is that the agents who’ll thrive are the ones who use AI matching as a starting point and then apply their own knowledge to refine the results.

Practically, this means:

Let the algorithm generate your long list. AI is better than any human at scanning a database of 500 buyers and identifying the 40 who are statistically most likely to be interested in your listing. Don’t fight this. It’s a genuine time-saver.

Apply your relationship knowledge to create the short list. Of those 40 algorithmically matched buyers, which ones have you spoken to recently? Which ones are motivated by something the algorithm can’t see? Which ones are actually ready to transact versus still “just looking”?

Use the gaps as prospecting opportunities. When the AI flags a buyer you haven’t heard from in three months, that’s a prompt to make a call. When it identifies someone searching in your area who isn’t in your CRM, that’s a lead to pursue. The technology works best when it’s feeding your human outreach, not replacing it.

The Vendor Conversation

This matters for listing presentations too. Vendors increasingly ask about buyer matching technology because they’ve read about it or seen it mentioned on portal marketing. You need to be able to explain what it does, how you use it, and — critically — what you add beyond what the algorithm provides.

The agents who position themselves as “I use AI to identify potential buyers and then I apply 25 years of local market knowledge to find the right one” are winning more listings than those who either ignore the technology entirely or claim it does everything.

Where This Goes Next

The buyer matching tools will keep improving. The datasets will get richer. The predictions will get more accurate. Within a few years, I expect these platforms will be able to predict buyer interest with enough accuracy that the “long list” step becomes almost automated.

But the human layer — the conversations, the relationship context, the emotional intelligence — isn’t going anywhere. If anything, it becomes more valuable as the data side gets commoditised. When every agent has access to the same AI matching tools, the differentiator goes back to who knows their buyers best.

That’s always been the job. The tools are just getting better at helping us do it.