AI for Australian real estate agents: listings and leads

How can real estate agents use AI in Australia?

Agents use AI for the writing and follow-up around a campaign, not for the licensed work. The useful workflows are vendor report drafts, listing copy drafts, a fast first reply and qualification for every enquiry, open-home follow-up, and rental enquiries with inspection bookings. Appraisals, price advice, negotiation and trust money stay with the licensed agent, and every public claim gets a human check before it goes out.

Workflow	What the AI does	What the agent does	Rule to watch
Vendor reports	Drafts the weekly update from campaign numbers and feedback notes	Writes the honest read and any price conversation	Underquoting
Listing copy	Drafts headline, description and feature bullets from notes and the floor plan	Checks every claim against the property	Australian Consumer Law
Enquiry reply	Replies within minutes, asks about finance and timeframe, books the inspection or call	Takes over the conversation	Privacy
Open-home follow-up	Same-day thank-you, a check-in, an alert to the agent on any reply	Calls the warm buyers	Spam Act
Rental enquiries	Answers routine questions, books inspections, sends reminders	Assesses applications and decides	Privacy, state tenancy law

Your CRM and portals stay the system of record. Australian agencies commonly run VaultRE, Rex, AgentBox, Agentpoint or Box+Dice, with enquiries arriving from realestate.com.au (REA), Domain and the agency website. Each differs in what outside tools can connect to, so check yours before choosing a design. Where no direct connection exists, the AI can work from the enquiry emails the portals already send you.

Can AI write vendor reports?

Yes, as a draft for the agent to finish. AI can turn the week’s campaign numbers (portal views and enquiries from your REA and Domain dashboards, open-home attendance, buyer feedback from your notes or CRM) into a plain-English weekly update in your agency’s voice. The agent writes or approves the part vendors care about most: the honest read on feedback, price and what to do next.

A template that keeps the AI to the facts:

Vendor: {name}
Property: {address}
Days on market: {N}
This week: {open-home attendance}, {portal views}, {enquiries}
Buyer feedback: {agent's notes, summarised, no buyer names}
Next week: {scheduled opens, advertising, actions}
Honest read: LEAVE BLANK FOR THE AGENT

Keep price out of the automated part. If campaign evidence moves the agent’s estimated selling price, Victoria and NSW treat updating it as a formal step for the agent (see the price rules below), not a sentence for the AI to improvise.

Is it legal to use AI-written listings or AI-edited photos?

Yes, with care. No Australian law bans AI-written listing copy or edited photos, but the agency is responsible for every claim in the ad, whoever or whatever produced it. Under the Australian Consumer Law, a false or misleading claim can breach the law even if nobody meant to mislead. NSW has also passed a specific disclosure rule for digitally altered or AI-generated rental listing images, expected to start in early 2027.

Listing copy: hard rules in the prompt

AI models drift towards real estate clichés and confident claims. Put these rules in the instructions every time:

You are drafting listing copy for {agency}, a real estate agency in {area}.
Use ONLY facts in the agent's notes, the floor plan and the photo list.
Never invent or infer: features, renovations, room sizes, views, distances,
school catchments, zoning, development or subdivision potential, rental
yield or grant eligibility. If a claim needs checking, write
[AGENT TO VERIFY] next to it.
Never write a price, price guide or price range. The agent supplies it.
Voice: {agency voice notes}. Australian English. No exclamation marks.
Output: a headline, a 200 to 300 word description, three feature bullets.

The agent reads every draft against the property before it goes to REA or Domain. The ACCC’s guidance on false or misleading claims applies to property ads like any other advertising, and our overview of the Australian AI compliance landscape covers the other regulators that touch AI use.

Photos, virtual staging and AI edits

An edit can mislead when it changes what a buyer or renter would see in person: removing a power line, a crack or a neighbouring building, or staging furniture that hides damage or makes rooms look bigger. In NSW, laws passed on 24/09/2026 will require landlords and agents to disclose when images in rental advertisements have been digitally altered or AI-generated to hide faults, with penalties up to $22,000, expected to start in early 2027 (NSW Government, 25/09/2026). The new rule covers rentals, not sale ads, but a misleading sale photo is still caught by the Australian Consumer Law. The safe habit in every state: label virtually staged or AI-edited images, and keep the unedited originals on file.

What price guide and underquoting rules apply to AI-drafted ads?

State law decides what price information an ad can carry, and the AI should never choose it. Victoria and NSW both have underquoting laws that tie advertised price information to the agent’s estimated selling price, and both cap price ranges at 10%. Some states also restrict price information for auction campaigns. The agent supplies approved price wording; the AI only places it, word for word.

As at October 2026:

• Victoria: an advertised price must be a single figure or a range of up to 10%, and must not be lower than the agent’s estimated selling price, the seller’s asking or reserve price, or a written offer the seller rejected as too low (Consumer Affairs Victoria).
• New South Wales: the estimated selling price goes in the agency agreement, a range can span no more than 10%, and the government’s guidance lists “$XXX,000+”, “offers over” and “offers above” as misleading terms agents must not use in advertisements (NSW Government).
• Other states and territories: check your regulator’s current rules before any template carries price text.

Can AI reply to and qualify property enquiries?

Yes. An AI lead system reads each portal, website or social media enquiry as it arrives, replies within minutes in your agency’s voice, asks what an agent would ask (buying or renting, finance position, timeframe, whether they have inspected), books an inspection or call, and logs it. It never gives price, finance or investment advice; a qualified lead goes straight to the agent with a short summary.

How we build it:

1. Map every enquiry source: REA, Domain, website forms, social lead forms and the shared inbox.
2. Write the routing rules with the principal: which agent gets which listing, and what counts as a hot lead.
3. Train the reply voice on 10–15 of your agents’ real replies.
4. Connect the calendar and CRM, or set up the email fallback where your CRM has no open connection.
5. Run drafts-only first: every reply lands for one-click approval until your agents trust it.

Can AI follow up open-home attendees?

Yes. A sequence can send each attendee a same-day thank-you with the listing details, a check-in a few days later, and an alert to the agent the moment anyone replies, so warm buyers get a phone call rather than a third email. Treat every one of these messages as a commercial message under the Spam Act: it needs consent, must identify your agency and must include a working unsubscribe.

When	Message	Sent to
Same afternoon	Thank-you, listing link, how to get the contract or ask a question	Everyone who signed in
2–3 days later	Short check-in: questions, a second inspection?	Everyone who has not replied
About a week later	Similar upcoming listings	Only attendees who agreed to marketing
Any reply	Sequence stops; agent alerted with a summary

Under the ACMA’s Spam Act rules, consent can be express or inferred, and unsubscribes must be actioned within five working days. The simplest protection is a clear, unticked marketing consent box on your digital sign-in, so the system knows who can receive the “similar listings” email.

Can AI book rental inspections, and what about tenant data?

Yes, for the routine part. An AI front desk on your website chat, SMS and email can answer questions about a listed rental from the ad, offer inspection times, book them and send reminders. It should ask only for what booking needs (name, contact details, preferred time) and never screen applicants. Assessing applications and every tenancy decision stay with the property manager.

Tenant data needs extra care. The same NSW reforms bar asking rental applicants for unnecessary details such as personal photos, social media accounts or details about their children, limit ID collection to the preferred applicant before the lease is signed, and add rules for storing and destroying data, with privacy penalties up to $11,000 for individuals and $49,500 for corporations (NSW Government). Keep identity documents and applications out of AI tools; our guide to what is safe to put into Claude or ChatGPT covers which plans suit customer data, and the AI receptionist guide covers how to test a front desk before tenants see it.

How On Autopilot can help

On Autopilot is a Melbourne AI consultancy run by Jenn Yang. For sales, the usual first build is the AI Front Desk ($1,500 setup + $199 a month, ex GST): it replies to portal and website enquiries on your listings within minutes, answers from the listing facts and books inspections, live in 7–10 working days including a shadow week. The AI Lead Engine ($2,000 setup + $499 a month) then follows up appraisal requests and post-open buyers and syncs to your CRM; we confirm on the audit call what your CRM allows. For property management, the same AI Front Desk answers rental enquiries by website chat, SMS and email and books inspections. Managed AI from $1,500 a month runs vendor report drafts, open-home follow-up and the rest. Month to month; you own the code. See AI for real estate sales agencies, or book a free 30-minute AI audit with Jenn.