Using AI to draft replies to tenant and applicant enquiries
How to get a clear, polite draft reply to a rental enquiry from an AI tool, and how to check it so nothing wrong or unfair goes out.
Enquiry replies are the easiest job to hand to a writing tool, and the easiest place to get caught out. Handling tenant enquiries is part of the day job. One general description of property management lists enquiries, applications, screening and leases together.1
The tool is good at tone. It's bad at facts. So you supply the facts and it supplies the manners.
Give it your facts, not its guesses
Never ask a tool "reply to this enquiry" and walk away. It will fill gaps with confident nonsense, such as an inspection time you never offered or a pet policy you don't have.
A tool can produce a hallucination, which is a response containing false or misleading information presented as fact.2 A made-up viewing time is exactly that.
Instead, write your notes first. Keep them short:
- The property is still available, or it isn't.
- Viewing times you've actually got free.
- How to apply, and what you need from applicants.
- What you can't say yet ("the owner hasn't decided on pets").
Then paste in the enquiry with the name and contact details removed, and ask for a short, polite reply using only your notes.
A prompt you can copy
Change the wording to suit you:
"Write a short, friendly reply to this rental enquiry. Use only the facts below. Don't add any times, prices or rules that aren't listed. If something is missing, say I'll get back to them. Facts: (your notes)."
That last instruction about missing details matters. It gives the tool a safe way out instead of inventing.
Where replies go wrong
Wrong facts. A draft can read well and still be wrong. AI outputs can often appear credible even when they contain errors or false information.3 Smooth writing isn't evidence of accuracy.
Unfair wording. The same Australian regulator warns that bias in training data may be replicated in a tool's outputs.3 In a rental reply that might show up as a tone that's warmer to one kind of applicant than another, or an assumption about a family or a job.
Read each reply and ask: would I write this to anyone who asked? If the answer's no, rewrite it.
Discriminating by accident. Rules about who you can turn away or how you can advertise vary by country. In the United States, the federal Fair Housing Act bans advertising a rental in a way that indicates preference, limitation or discrimination based on race, color, religion, sex, disability, familial status or national origin.4 That's a US rule. Australia, the United Kingdom and other places have their own, so check your own regulator.
Don't ask a tool whether a particular applicant "seems suitable". Selection is your call, made on your written criteria, not a chatbot's.
Keep it short
Applicants want three things: is it available, when can I see it, how do I apply. A good reply answers those in a few lines. If the draft runs long, cut it.
Cut these as well:
- Gushing openers and sales talk about the "perfect" home.
- Promises about repairs or timing you can't back.
- Anything about the owner's plans you haven't confirmed.
Check before you send
Read it against the listing and your calendar. Check:
- Address and unit match the right property.
- Dates and days of the week are right.
- The application steps match your actual process.
- There's no promise you haven't cleared with the owner.
Then send it from your own account. The tool doesn't send anything; you do.
One more habit: build a few of your own approved replies (available, not available, how to apply) and keep them in a folder. The tool then only has to adjust small details, and you review less each time. For wider rules on what the tool should never see, read what not to paste.
Cited material
- Property management, Wikipedia
- Hallucination (artificial intelligence), Wikipedia
- Guidance on privacy and the use of commercially available AI products, Office of the Australian Information Commissioner (OAIC)
- Fair Housing Act, Wikipedia