Nov 22, 2024
More Sydney buyers are uploading contracts of sale, strata reports and building and pest reports into general AI chatbots such as ChatGPT, then asking whether anything should worry them. A tidy, reassuring summary arrives in seconds. With auction campaigns running only a few weeks and pre-auction reports that can stretch past 100 pages, the appeal is easy to understand.
Used carefully, AI can give buyers a faster first read and a sharper list of questions for their conveyancer. Used carelessly, it can create false confidence right before an unconditional bid. In NSW that often means auction day, when there is no cooling-off period to fall back on.
The biggest risk is a confident answer that is wrong. General chatbots can “hallucinate”, inventing a clause, a levy figure or a date that sounds entirely plausible. They deliver a mistake in the same assured tone as a fact, and unless the tool shows which page an answer came from, the only way to check it is to reread the document yourself.
Gaps are the second risk. Scanned pages, handwritten strata minutes, tables and annexures are often only partly read, and a summary built from 60 pages of a 90-page report still looks complete. The issue that matters, such as a remedial works quote buried in old committee minutes, is often sitting in exactly the pages that were skipped.
General tools can also mix up state rules. Disclosure obligations, cooling-off rights and strata law differ from state to state, and a generic answer can apply Victorian rules to a NSW contract. In NSW, a private treaty purchase generally comes with a five-business-day cooling-off period, while a purchase at auction comes with none.
Finally, a chatbot only sees what it is given. It does not know about a development application lodged next door, a flood overlay on the planning certificate, or what comparable units in the same block sold for last month. Buyers should also check where a consumer tool stores their documents, which carry names, addresses and financial details, and whether it uses them for training.
Most of these problems come from using general-purpose software for a specialist job. Chatbots such as ChatGPT are built to answer anything, from recipes to work emails, and were never designed to read a NSW strata report, interpret a planning certificate or show where each answer came from. Purpose-built property software is designed around those documents and rules, and buyers who want to use AI for due diligence should choose a tool made for the job over a general chatbot.
The useful question is not whether to use AI, but whether its answers can be checked. A trustworthy tool cites a source for every answer, so a buyer can go from “there is a $48,000 special levy” straight to the page of the strata report that says so. It reads the whole document, including scanned pages and the minutes at the back, and it says nothing was found rather than guessing when the documents are silent.
It should also know where the property is. That means applying NSW rules and terminology to a NSW purchase, and checking the zoning, overlays and nearby development for that specific address rather than answering in general terms.
Most importantly, a professional still signs off. A conveyancer or solicitor, a licensed building inspector and a buyers agent remain accountable for the advice a buyer relies on. AI should do the reading and leave the judgement to the people who answer for it.
Pivot Property Buyers has partnered with Moving Castle, an Australian AI workspace for property professionals. It is currently in beta, but it is being developed by a genuinely world-leading technical team.
Moving Castle is built around the tests above rather than general chat. Each answer about special conditions, levies or defects links to the exact page of the contract, strata report or building and pest report it came from, so findings can be checked rather than simply trusted. It reads the whole pack, scanned pages included, so the issue buried in old committee minutes is found too.
It also researches beyond the documents, looking at zoning, overlays, nearby development applications and comparable sales, which is the context a chatbot never sees. Every property, client brief and document for a purchase sits in one place, from the first inspection through to auction day.
In practice, the platform acts as a second set of eyes, surfacing questions quickly so Pivot’s team can spend more of its time on judgement. That might be whether a special levy changes a buyer’s bidding limit, whether a clause is worth negotiating, or whether a property should come off the shortlist altogether.
Because the platform is still in beta, Pivot Property Buyers continues to run its full due diligence process manually, in parallel. Every contract, strata report and building and pest report is still read by the team, and Moving Castle’s findings are checked against that manual review rather than replacing it. Those findings also go to the buyer’s conveyancer and building inspector, never in place of them.
AI will not replace due diligence, but it has real capacity to enhance it. Used well, it brings important questions to the surface sooner and makes it harder for a costly clause or levy to slip through unnoticed. That is good news for buyers, provided every answer can be checked and a professional stands behind the advice.
Buyers doing it themselves should use AI to find questions rather than answers, check every important claim against the page it came from, and never go unconditional on the strength of a chatbot summary. A conveyancer or solicitor should still review the contract, and a licensed building and pest inspection should be booked well before auction day.
For buyers who would rather have it done properly, Pivot Property Buyers combines Moving Castle’s research with deep local knowledge across Sydney. Call 1300 402 424 or visit the office at 197 Clovelly Road, Clovelly, to talk through due diligence on your next purchase before the campaign begins.
This article is general information only and is not legal or financial advice.
Nov 22, 2024
Nov 22, 2024
