AI in Malaysia · Property agencies

One property, ten listings: what AI should fix for a Malaysian property agency, and what it could make worse

Up to 40% of subsale listings online are estimated to be duplicates, outdated or dummies, and rental fraud cases tripled in 2025. AI that writes listings faster adds to the pile. The useful kind answers enquiries, keeps the listing book true and handles the paperwork.

Pikabyte Engineering 7 min read

30–40%

of subsale listings on Malaysian property portals are estimated to be duplicate, outdated or dummy posts, according to industry data experts cited by The Star

A survey of Malaysian property portals found around 450,000 to 500,000 active online listings, The Star reported in 2026. Industry data experts cited in the same report estimate that 30% to 40% of the subsale listings are duplicate posts, outdated posts or deliberate dummy posts placed to collect enquiries.

The market behind those listings is large. Malaysia recorded 416,413 property transactions worth RM241.87 billion in 2025, the highest value in a decade. The first half of 2026 brought another 187,320 transactions worth RM105.12 billion, and 59.3% of them were residential.

The profession is large too. According to Board figures cited in The Star in November 2026, Malaysia has more than 56,000 registered real estate negotiators, over 4,000 probationary estate agents and more than 3,000 registered estate agents.

So the ordinary state of the business is many people, many listings and a buyer who cannot tell which advertisement is real. That is the problem to point AI at.

Writing listings faster is not the answer

The portals already do it. PropertyGuru’s site for Malaysian agents advertises “AI that writes for you and creates videos that bring your listings to life”. A negotiator can produce a polished description in seconds.

Descriptions were never the scarce thing. The Star’s report describes sellers handing keys, photos and floor plans to as many as twenty agents at once, and agents who close other deals and never update or delete the original post. A tool that makes posting cheaper produces more of the same.

The professional standard is already clear. Standard 6 of the Malaysian Estate Agency Standards does not allow misleading advertisements. As The Star summarises it, a public listing must represent the true asking price and be authorised in writing by the client.

The buyer no longer trusts the advertisement

Police recorded 184 rental fraud cases in 2023, 307 in 2024 and 922 in 2025, when losses reached about RM2.5 million. The director of the Commercial Crime Investigation Department described fake profiles posing as agents or owners, unusually low rents, and pressure to pay a deposit before any viewing.

The Board’s registrar said in October 2024 that about 1,500 people had lost nearly RM60 million to unlicensed brokers over five years. In 2026 The Star reported scammers using AI-generated images and photos stolen from other listings. One victim checked the negotiator’s number. It was real, but it had expired.

Acting as an estate agent without registration is an offence under the Valuers, Appraisers, Estate Agents and Property Managers Act 1981, with a fine of up to RM300,000, up to three years in prison, or both. In August 2025 PropertyGuru introduced a Verified badge that requires an identity check, licence validation and contact confirmation.

For a legitimate agency this is an opening. Being easy to verify is now part of the service.

Where AI is useful today

1. Answering the enquiry first, and saying who is answering. Portal and advertisement leads end up as WhatsApp chats at all hours. An assistant can answer from the listing record: price, size, tenure, maintenance fee, whether it is still available. It can ask about budget, area, timing and financing, book the viewing and pass the chat to the negotiator. It should say that it is an assistant, and give the negotiator’s name and registration number in the first reply. We set out Meta’s rules and costs in AI agents on WhatsApp. AllSet, the platform we are building with Northridge Technologies of Singapore, is in development for businesses of this kind.

2. Keeping the listing book true. This is the job that separates a trusted agency from the pile. Software finds the same unit listed by three negotiators in your own firm. It flags a listing nobody has touched in thirty days and asks the owner whether the unit is still available and at what price. A model reads the owner’s reply and proposes the update. The system checks that a written authorisation is on file before anything is published.

3. Writing the advertisement from the record. Once a listing is a set of confirmed facts, a model can write it in English, Malay and Chinese for each portal. It writes only what is in the record. Built-up area and tenure come from the documents, not from memory.

4. Reading the documents. A title search, a sale and purchase agreement, a tenancy agreement, an assessment bill. A model can pull out the tenure, the land area, the registered owner and the conditions into a one-page fact sheet for the negotiator to confirm. The conveyancing remains the lawyer’s work.

5. Customer due diligence. Registered estate agents are reporting institutions under the anti-money laundering law. Bank Negara Malaysia’s policy document, in effect since 6 February 2024, requires them to conduct customer due diligence on both purchaser and seller, or landlord and tenant, and to keep records for at least six years. It treats negotiators as part of the firm. Reading identity documents into the form, checking that it is complete and filing it against the deal is routine work a system can do. Deciding that something looks suspicious is not.

6. Keeping track of both sides. An agent carries a seller’s price and a buyer’s budget back and forth, message by message. A model can summarise where each side stands and draft a neutral version of an offer. The negotiator sends it.

What it should not do

Invent a fact or a price. The asking price is the owner’s written instruction. A model that fills a gap with a plausible number has published a misleading advertisement in your name.

Dress up the photographs without saying so. We could not find a Malaysian rule written specifically for AI-staged listing photos. The existing standard on misleading advertisements covers it. Label a staged image as staged, and never remove a crack, a pylon or the neighbouring block.

Value a property. A model’s estimate from comparable listings is a talking point. Valuation is a registered profession under the same Act.

Message strangers. Under the Personal Data Protection Act (PDPA), a person can require you in writing to stop using their data for direct marketing. A purchased list plus a tireless AI assistant is a quick route to a complaint to the Board or the Commissioner.

Screen tenants or buyers on its own. Guidelines issued by the Personal Data Protection Commissioner on 30 April 2026 say AI should not be relied upon as the sole basis for a decision about a person. Law firms describe the guidelines as not legally binding, but as the regulator’s expectation.

Where the client’s data goes

An agency holds identity cards, payslips, bank statements and tenancy histories.

Juwai IQI told StarProperty in September 2025 that it runs open-source models on its own servers for customer service chatbots, PDF summarisation and marketing content, which keeps customer data inside the company. A twenty-person agency does not need its own servers. It does need to know where its AI provider processes documents and how long they are kept. We explain how to check in where your AI runs.

The larger risk is quieter. Negotiators work on their own phones, and some will paste a client’s documents into a free chatbot to save time. Give them an approved tool and a one-page policy.

A sensible first project

  1. Put every listing in one place, with the owner’s written authorisation, the confirmed facts and the date it was last checked.
  2. Remove the duplicates and the dead listings. This costs nothing and improves every later step.
  3. Let an assistant answer first enquiries from that record, in draft mode for two weeks, with negotiators pressing send.
  4. Add the due diligence checklist to every deal file.
  5. Measure three things: time to first reply, share of enquiries that became viewings, and listings older than thirty days.

The agencies that gain from AI will be the ones whose listings are real and current, and whose negotiators a buyer can verify in a minute.

Sources

This article is general information and is not legal advice.

Written by

Pikabyte Engineering

Pikabyte builds custom business systems, applied AI and secure cloud infrastructure from Malaysia. Talk to us.

The work behind this article

Platform · AI agents · WhatsApp commerce

In development

A team of AI staff that answers, books, sells and chases payment on WhatsApp, with people in charge.

A multi-tenant platform for small and medium businesses whose customers reach them on WhatsApp. AI agents answer every chat from the business's own knowledge, book appointments, score leads and follow up on invoices, and hand over to a person when one is needed. Behind the inbox sits the system that runs the business: CRM, calendar and rosters, jobs and quotations, stock, invoicing and automation.

8
AI staff roles, from front desk to owner's assistant
589
automated backend tests
5
integrations: Google, Shopify, TikTok, Lalamove, Do Not Call registry
1
gateway every AI call passes through, metered for cost
Tour AllSet Northridge Technologies · Singapore

Start a conversation

Want this applied to your business?

Tell us what is slowing the business down. You will talk to an engineer, not a salesperson, and get a straight answer on whether software is the right fix.