By mid-2026 most Australian agents have given up the “should we use AI for listing copy” debate. The question is which tool, used how, and for which kind of campaign. Three patterns have settled out across the agencies we talk to in Sydney, the Inner West and the Central Coast.

The first is the platform-native tools. REA and Domain both ship listing-copy generators now, and they are getting used. Their advantage is that they pull straight from the property record and produce copy that lands on-brand with the platform style. Their weakness is sameness. Across a suburb you can spot a REA-generated description by lunchtime on a Saturday opens day. Agents who care about differentiation are using the platform tools as a first draft and rewriting the lede.

The second is the agency-owned AI assistant. Bigger franchise networks have rolled out internal tools trained on their own past listings. These are the most consistent. They pick up house-style on adjectives, lengths, and how to handle awkward features like a tandem block or a no-stamp split-level. The output reads like the agency, not like a generic large model. The trade-off is the cost of getting them set up and the constant feed of feedback to keep them tuned.

The third is the agent’s own ChatGPT or Claude workflow. This is now the dominant pattern for solo principals and small offices. It is also the pattern where most of the bad copy comes from. Agents writing prompts on the fly produce filler. Agents working from a documented prompt template — house style, must-include features, banned words, tone — produce copy that holds up against the platform tools.

The piece of work nobody seems to want is the editorial review. AI copy hallucinates features. It writes about a “sunny north-facing terrace” when the orientation is the other way. It invents a study nook. It rounds a 412 m² block to “more than 450 m².” The agencies getting the most lift from AI listing copy are the ones treating the AI as a junior copywriter and the agent as the editor.

For agencies thinking about going further — agent triage, vendor communications, automated callbacks — the work moves from a copy tool decision to an operations decision. That is the bit where it pays to talk to an AI implementation partner. Team400 is one of the AI consultancies in Sydney working on the back-office side of agency operations, and they are a reasonable conversation if you are thinking past listing copy.

Listing copy is the easy part of AI in real estate. The hard work is everything that happens after the call gets booked.