AI, Tech & Innovation

AI Skills Series, Episode 1: How to set an AI policy for your real estate agency

AI Skills Series, Episode 1: How to set an AI policy for your real estate agency 
Back to blog

Your team is already using AI. The question is whether they’re doing it in a way you’d be comfortable explaining to a vendor, a landlord, a tenant or the privacy regulator. 

On the sales desk, Agents are writing listing copy, appraisal letters and buyer follow-ups in ChatGPT. In property management, PMs are drafting routine inspection reports, rewriting arrears letters and pasting lease clauses into whatever tool is open in the next browser tab. Most of it is well-intentioned and plenty of it saves real time. But every one of those copy-and-paste moments moves vendor, buyer, tenant and owner information into a system your agency hasn’t assessed, under terms you probably haven’t read. 

An AI policy doesn’t need to be a 30-page legal document. For most agencies it’s two or three pages that answer four questions: which tools are approved, what information can go into them, who checks the output and who owns the policy. This guide walks through how to set one, and why the way your team connects to AI matters as much as which AI they use. 

What’s actually at risk 

The biggest risk isn’t AI itself. It’s personal information leaving your control without anyone deciding it should. 

Privacy obligations. In Australia, the Office of the Australian Information Commissioner (OAIC) is direct about this. Its guidance on commercially available AI products recommends organisations “do not enter personal information, and particularly sensitive information, into publicly available generative AI tools.” Entering it can count as a disclosure to the tool’s provider under APP 6. 

Don’t assume the small business exemption covers you. Agencies under $3 million turnover are generally exempt, but REINSW notes that keeping a residential tenancy database, or providing property management services, is likely to bring an agency back under the Act. And if your privacy policy makes promises, you’re bound by them regardless. 

In New Zealand there’s no size threshold. The Privacy Commissioner has said its AI guidance is “for all agencies using AI tools in New Zealand”, and recommends a privacy impact assessment before you start. 

Accuracy. The OAIC notes that “AI systems are known to produce inaccurate or false results”. A made-up rent figure in an owner report, or a wrong date in a breach notice, is your agency’s error, not the software’s. 

Client trust. Vendors share their reserve, their reasons for selling and their financial position. Landlords hand you their largest asset, and tenants hand you their ID, income and rental history. If any of them found their details had been pasted into a personal chatbot account, the reputational cost would outweigh any time saved. 

Sales-specific exposure. Sales teams carry risks property management doesn’t. A vendor’s reserve or settlement terms pasted into a chatbot is confidential information you’ve lost control of. AI-written listing copy that overstates a property, or an AI-enhanced photo that hides a defect, can still be misleading conduct under the Australian Consumer Law, whoever wrote it. NSW has already passed rules requiring disclosure of AI-generated or digitally altered images in rental ads, expected to start in early 2027, and sales ads are likely to face the same scrutiny. 

Identity documents under AML/CTF. Since 1 July 2026, agencies providing services regulated under AUSTRAC’s tranche 2 reforms must verify customer identity and keep records for seven years. That makes ID handling a compliance process with its own rules, and those documents have no place in a general AI tool. 

Shadow use. Banning AI outright rarely works. It pushes use onto personal accounts and phones, where you have no visibility at all. 

Six steps to setting your policy 

A workable policy takes a Principal and one or two senior staff about a fortnight. These six steps cover what it needs. 

1. Find out what’s already happening 

Ask both your sales and property management teams, without judgement, which AI tools they use and what for. Sales Agents and PMs often use AI in quite different ways, so ask each group separately. You’ll likely find a mix of free personal accounts, browser extensions and AI features built into software you already pay for. You can’t write sensible rules until you know the starting point. 

2. Decide which tools are approved 

Pick a short list of approved tools and require business accounts, not personal ones. Business and team plans from the major AI providers generally give you admin controls and clearer commitments on whether your data is used for model training. Check the terms yourself before approving anything. The OAIC expects organisations to “conduct due diligence to ensure the product is suitable to its intended uses”. 

3. Set clear data rules 

This is the core of the policy. Group information into three tiers so staff don’t have to make judgement calls on the fly: 

  • Never: identity documents and AML/CTF records, bank details, tenancy application financials, health information, a vendor’s reserve or confidential sale terms and anything about a dispute or tribunal matter. 
  • Only through an approved, connected or built-in tool: tenant, owner, vendor and buyer records, rent and arrears data, inspection and maintenance history, appraisal and campaign data. 
  • Fine in any approved tool: generic templates, listing copy drafted from public property features, marketing copy without client details, general research and training questions. 

4. Keep a human accountable for every output 

Anything that goes to a client, a tribunal or a trust account record must be read and approved by the person sending it. For sales, that includes listing copy, price guides, appraisals and any edited listing photos: the Agent signs off that every claim is accurate and every image fairly represents the property. AI drafts; your people decide. Write that down plainly. 

5. Be open with clients 

The OAIC recommends updating privacy policies and collection notices with “clear and transparent information about their use of AI”. A short paragraph in your privacy policy and tenancy application forms is usually enough. 

6. Train, review and assign an owner 

Name one person who owns the policy. Run a 30-minute session when it launches and add it to onboarding. Review it at least every six months, because the tools change faster than most agency policies do. 

Why connected AI is safer than copy and paste 

How your team gets data into AI matters as much as which AI they use. There are two broad ways to do it. 

The first is the browser approach. A Property Manager exports a report or copies records from your property management software, then pastes them into a chatbot. Some browser extensions go further and read/ write whatever is on screen. 

The second is a connected approach using the Model Context Protocol (MCP), an open standard that lets an AI assistant query business software directly through an approved connection. Instead of moving data around by hand, the AI asks your system for what it needs, under the same login and permissions the user already has. 

 Copy and paste or browser extension Connected via MCP 
Who decides what data is shared Each staff member, case by case Your system’s existing user permissions 
How much data leaves at once Whole exports, often more than the task needs What the question needs, within the user’s role 
Turning off access Hard; copies already sit in chat histories Revoke the user’s connector access 
Data copied into files and chats Often, as spreadsheets and pasted text No exported files to lose or forward 
Accuracy Stale the moment it’s exported Live data at the time of the question 
Visibility for the principal Close to none A defined, approved channel 

The practical difference is control. With copy and paste, your data policy depends on every staff member making the right call every time. With a connector, access is decided once, by you, and enforced by the system. 

Where connected AI still needs care 

MCP is not a privacy exemption. A few things still apply: 

  • The AI account matters. Data retrieved through a connector still goes to the AI provider. Use business accounts and check their data retention and training settings. 
  • Only use official connectors. Unofficial MCP servers built by third parties can see whatever you grant them. Stick to connectors published by your software vendor. 
  • Outputs are still your responsibility. A connector gives the AI accurate inputs. It doesn’t guarantee accurate answers, so step 4 still applies. 
  • Grant access deliberately. Start with the staff who need it, not the whole office. 

How PropertyMe’s AI tools fit your policy 

If you use PropertyMe, AiMe Connect gives your property management team the connected option described above. It launched on 8 October 2026 and uses MCP to link your agency’s own Claude or ChatGPT account to live PropertyMe data. 

Staff can ask questions in plain language, such as which leases are due for renewal next month or which maintenance jobs have been open longest. The AI can draw on properties, tenancies, inspections, maintenance jobs, tasks, lease renewals, team members and contacts and return answers as tables, charts or spreadsheets. 

A few details matter for your policy: 

  • It’s read-only. At launch the AI can look up data but can’t change anything in your portfolio. Any action still needs to be prompted by a human and happens in PropertyMe. 
  • You choose who gets access. The portfolio subscriber grants MCP access to individual team members in Portfolio Settings, and the connection follows each user’s existing PropertyMe permissions. 
  • Each user signs in and consents. Staff log in with their own PropertyMe account and approve access before the AI can see anything. 
  • It’s available on Manage PM Advanced and above, in Claude and ChatGPT on desktop, web and mobile. 

In policy terms, AiMe Connect is the tool for your middle data tier: the client information staff need AI help with but shouldn’t be pasting anywhere. The setup guide covers the steps. As PropertyMe itself recommends, review the data and privacy settings of whichever AI account you connect. 

Getting help with setup 

If you’d rather not set this up on your own, PropertyMe partners Sidekick and The Gallery AI can get you started. The Gallery AI sets up your agency’s Claude or ChatGPT business accounts, switches on AiMe Connect for the right people and tests each connection before your team relies on it. Then it trains staff to use it day to day. Sidekick coaches property management business owners so you can lead AI adoption yourself. Both start with a free 20-minute call. Register your interest. 

For sales teams 

AiMe Connect currently covers property management data. For sales, the equivalent safe route is AI built into the CRM your Agents already use. PropertyMe’s Grow CRM includes AI features for composing emails and SMS, suggesting next steps and following up with contacts. Because those features work inside the CRM, Agents don’t need to copy buyer or vendor details into a separate chatbot to get the same help. 

Whatever CRM you use, the policy principle is the same: point your sales team to AI that works inside your approved systems, and keep confidential sale information out of standalone chat tools. 

Your AI policy checklist 

Copy these headings into a document and fill in each one for your agency. Most agencies can fit the finished policy on two or three pages. 

  • Purpose and scope. Who the policy covers (staff, contractors, virtual assistants) and which work it applies to. 
  • Policy owner. The person responsible for the policy, approving new tools and answering questions. 
  • Approved tools. The list of AI tools staff may use, and the requirement to use business accounts only. 
  • Connected tools. Which integrations (for example AiMe Connect) are approved, and who has been granted access. 
  • Data rules. Your three tiers: never share, connected tools only, fine in any approved tool. 
  • Prohibited uses. For example: tenant screening decisions made by AI alone, uploading ID documents, using AI for trust accounting entries, presenting an AI price estimate as an agent’s appraisal, and publishing AI-altered listing images without disclosure. 
  • Human review. Who must check AI output before it goes to clients, tribunals or into records, including listing copy, price guides and edited photos. 
  • Client transparency. The wording added to your privacy policy and application forms. 
  • Incidents. What staff do if personal information is shared in breach of the policy, and who they tell. 
  • Training. When staff are trained and how the policy is included in onboarding. 
  • Review date. When the policy will next be reviewed (at least every six months). 

This checklist is a starting point, not legal advice. Have your privacy obligations reviewed by a lawyer or your state’s real estate institute, particularly if you’re unsure whether the small business exemption applies. 

Start with the data rules 

If you only do one thing this month, write down what information can and can’t go into AI, and tell your team. That single page closes the biggest gap. 

Then give staff a safer way to get the help they’re already looking for. Banning copy and paste only works when there’s a better option sitting next to it. For property management teams on PropertyMe Manage PM Advanced, read the AiMe Connect setup guide to see how it fits your policy. For sales teams, start with the AI features already in your CRM. 

Sources 

Disclaimer: This article is general information only and is not legal advice. Privacy, AML/CTF and advertising obligations differ between Australian states and territories, New Zealand, and the services your agency provides. Get advice from a qualified legal professional before making decisions about your agency’s AI policy or compliance obligations. 

Become a PropertyMe Monthly Insider

Join 65,000+ Property Managers, Real Estate Agents and Principals. Get the latest industry news and actionable tips in your inbox.