AI for Rental Properties: Screening Enquiries, Viewings and Tenant Replies

A landlord with several rental properties can use AI to draft replies to enquiries from a short fact sheet per property, collect the same screening answers from every enquirer, offer fixed viewing slots from a calendar, and draft routine tenant replies for approval. Keep tenant selection, repairs and anything legal in human hands.

The gain is mostly in the first day or two after an advert goes live, when enquiries bunch up and most of them ask the same five questions. The catch is fairness: a system that replies faster but asks different people different things, or sorts them by impression, creates a fair-housing problem that speed cannot make up for. So the same questions go to everyone, and a person makes every decision.

Follow me on Instagram@sagnikteaches

The landlord and the problem

An illustration. A landlord owns six rental homes, two of them flats, four houses, with one or two properties empty at any time. Each advert brings roughly 25 enquiries in the first week, mostly through a listing site's message box and a few by text. The landlord has a day job and answers in the evenings, usually 12 to 36 hours after the message arrived. By then, a third of the enquirers have booked viewings elsewhere.

Connect on LinkedInSagnik Bhattacharya

The time cost is about 12 minutes per enquiry across reading, answering questions already in the advert, asking about move-in dates, and arranging a viewing by text, with typically three messages back and forth. At 25 enquiries per vacancy that is five hours, and with about eight vacancies a year that is 40 hours a year on the first reply alone. Add viewings (an hour each including travel), tenant maintenance requests and rent questions, and the evenings fill up. These are the assumptions to replace with your own counts; the arithmetic is the point.

Subscribe on YouTube@sagnikteaches

Start with a fact sheet for every property

AI replies are only as good as the facts you give it. Build one short fact sheet per property, kept in a project folder in ChatGPT or Claude (both now include projects even on their free plans; Claude's free plan caps them at five) or in a single document you paste into each chat. An illustrative sheet:

Property ref: FLAT-2B
Type: 2-bed flat, second floor, no lift
Rent: $1,450 a month; deposit: one month and a half
Available from: 1 December; minimum term 12 months
Bills: tenant pays electricity and internet; water included
Pets: considered, case by case, one small pet, extra $25 a month
Parking: one allocated space
Smoking: not allowed
Viewing slots: Tuesday and Thursday 6 pm to 8 pm, Saturday 10 am to 1 pm
Application: enquirers send the details in the form; no payment is taken before a viewing
Do not say: anything about the neighbours, the area's crime or schools

The last line is a fair-housing safeguard. Property descriptions and replies should not steer people with comments on who lives nearby or who would "fit in". Keep the sheet to facts about the property and the terms.

The enquiry reply: one prompt, same questions for everyone

The prompt below drafts a reply to any enquiry. Paste the enquirer's message and the fact sheet, and get back a draft to approve.

You are replying on behalf of a small landlord. Use only the property fact
sheet. Write a friendly reply under 120 words that: answers any question
asked using the sheet; says if something is not on the sheet and that I
will confirm; asks the same four questions every time (move-in date,
number of people who would live there, whether there are pets, and
whether they can share proof of income at the viewing stage); and offers the
next two viewing slots I give you. Never ask about or comment on age, family
status, religion, ethnicity, disability, nationality or any protected
characteristic. Do not promise the property or give legal advice.

An illustrative enquiry and draft: the enquirer writes, "Hi, is the flat still available and are cats allowed?". The draft returns: "Hi [first name], yes, FLAT-2B is available from 1 December on a 12-month term at $1,450 a month. Pets are considered case by case, with one small pet and an extra $25 a month. Could you tell me your ideal move-in date, how many people would live there, whether you have any pets, and whether you could bring proof of income to a viewing? I can offer Tuesday at 6 pm or Saturday at 10 am. Thanks, [your name]". You change nothing or one word, then send. Time: under two minutes instead of twelve.

Check the draft for two common errors. First, the model may state a figure that is not on the sheet, such as a deposit, because it sounds plausible. Second, it may reassure beyond your sheet ("a quiet building"). Both are fixed by the "use only the sheet" line and by reading before sending.

Screening enquiries fairly

Screening at the enquiry stage should mean sorting for practical fit, not judging people. A fair method has three parts.

1. Written criteria, set before the advert goes up. For example: monthly income at least 2.5 times the rent shown by documents, a stated minimum term, pet rules, occupancy limits that reflect the size of the home, and a referencing check on every applicant who reaches the application stage. Equal treatment is the point: the same criteria, the same questions, the same order.
2. The same four questions to everyone, as in the prompt above. The AI collects answers; it does not rank people.
3. A neutral sort. Ask the model to sort enquiries into "meets stated criteria", "missing information" and "outside stated criteria" with a one-line reason tied to a criterion, and to leave out anything about the person other than the stated criteria. You then read the "outside criteria" group, because model sorts can be wrong.

What to avoid: asking an AI to "pick the best tenant", to infer anything from a name or a writing style, or to summarise social media. Fair-housing and anti-discrimination rules differ between places, as do any exemptions for small landlords, so check the rules that apply to your properties with a local adviser or your landlord association and write down the screening criteria. Automated tools can make discriminatory patterns harder to see, not easier, so keep a record of who enquired, who was offered viewings and why, and look at it each quarter.

Scheduling viewings without the back-and-forth

Viewings are where messages multiply. A booking page removes most of the traffic: a free calendar tool with appointment slots (Google Calendar, for one, lets a free personal account publish a single booking page; check what yours includes) lets enquirers choose a time inside the windows you set. The AI draft then ends with the booking link rather than two slot suggestions.

Three rules make viewings run well:

• Group viewings into windows. Two or three viewings of the same property within 90 minutes are easier than six single trips. Offer the same window to everyone.
• Confirm and remind. A confirmation straight after booking and a reminder the evening before help cut no-shows. Check which of these your booking tool sends by itself: Google Calendar keeps automatic email reminders for its paid plans, for example. Ask the AI to write both messages from a template, and if the tool cannot send the reminder, send it yourself from a saved text.
• Accessibility. Include a line saying whom to contact if the viewing needs adjustments, such as step-free access or a different time. Handle those requests personally and promptly.

An illustrative template for the reminder: "Hi [first name], a reminder that your viewing of FLAT-2B is tomorrow at 6.20 pm. Please bring photo ID and, if you'd like to apply, proof of income. If you can no longer make it, use the link to cancel so someone else can have the slot. [your name]". After the viewing, a follow-up with the application link and the closing date goes to everyone who attended.

Replying to tenants: drafts for you to approve

Current tenants write about repairs, rent dates, parking, noisy neighbours and move-out. Here AI is a drafting aid, and rules matter more than speed.

Routine messages worth drafting

Rent reminders, bin-day or access notices, thank-you notes after a repair, and answers to factual questions in the tenancy agreement ("When is rent due and by what method?"). Keep a short document of standard answers and let the AI personalise them.

Messages to read twice

Anything about a repair affecting health or safety, damp, heating failure in winter, a leak, a lock, deposits, rent rises, notices, arrears or complaints. The AI can draft a calm first response, but you add the time commitments and legal wording, and you decide who attends. An illustrative before and after: the tenant writes angrily that the boiler has been down for two days. A bad draft says "We will look into it soon". A good draft, edited by you, says "I'm sorry about the heating. I've asked the engineer to call on Wednesday morning; if you have not heard from them by 10 am, please text me. In the meantime I can arrange a heater, tell me if you need one." The AI makes it kinder, you make it accurate.

A maintenance triage prompt

Classify this tenant message as: emergency (risk to safety or health, no
heating or water, security), urgent (affects daily use), routine, or
question. Give one sentence of reasoning and a suggested first action for
the landlord. Do not diagnose the fault. If unsure, choose the higher
category.

"Choose the higher category if unsure" matters because the costly error is rating an emergency as routine. Check the labels for the first month and tell the model where it got one wrong.

The month in numbers

Back to the six-property landlord. In the first month with the system, one flat goes live and attracts 28 enquiries. All 28 receive a reply the same day, 25 of them within two hours, because the landlord approves drafts from a phone at lunch and in short breaks (28 drafts at about two minutes each is under an hour in total, against about five and a half hours before). Twenty-one answer the four questions; 15 meet the written criteria; the landlord offers viewing slots to those 15 and also to two enquirers who were missing proof of income but met everything else, because the criteria allow it at the application stage. Nine book through the calendar link, and eight attend one of two viewing windows. Six apply and the landlord references the two best matches against the written criteria.

The saving on the first reply and scheduling is around four to five hours for that vacancy, and the viewing windows save a further two to three hours of travel. On the cost side: an individual paid chat plan at $20 a month on monthly billing (ChatGPT Plus and Claude Pro both list that figure; check the current price), a free calendar booking page, and about an hour to write the fact sheet and prompts. Over a year of eight vacancies, a saving of roughly 40 hours against about $240 of subscriptions is plausible, but all these figures are assumptions. Measure yours for two vacancies before you believe them.

Three more messages, worked through

A vague enquiry that needs one clarifying question

Some enquiries say only "interested, call me". The draft should not ask a dozen questions. A better reply gives the headline facts and a single easy next step: "Hi [first name], thanks for getting in touch. FLAT-2B is $1,450 a month from 1 December. The quickest way forward is to pick a viewing slot here [link] and tell me your preferred move-in date. If a call suits you better, tell me a good time." Ask the model for a short version and a longer version of this kind of reply, then keep both in your document of standard answers.

A polite decline that stays consistent

After viewings you will turn people down. Use one template for everyone, tied to the written criteria and nothing else. An illustrative version: "Hi [first name], thank you for viewing FLAT-2B and applying. I have decided to go ahead with another applicant whose application met all of the stated criteria at this stage, so I will not be taking yours forward this time. I appreciate your time and wish you luck with your search. [your name]". Have the AI polish wording but never add reasons you have not written down. If an applicant asks why, answer from the criteria, in writing, and keep the record.

A rent-late first message

A first reminder should be factual and calm, with no threats. Draft: "Hi [first name], I haven't yet seen this month's rent of $1,450, which was due on the 1st. Could you let me know when to expect it? If something has gone wrong with the payment, tell me and we can sort it out. [your name]". Anything after the first reminder, such as formal notices, payment plans or legal steps, is not for AI to draft unsupervised: use a template from your landlord association or a solicitor, and have a person decide.

A weekly five-minute review

A system like this drifts unless someone looks at it. Each Friday, spend five minutes on a small table, kept in a spreadsheet:

The last two columns tell you whether to trust the drafts more or less. If "drafts edited" stays high, the fact sheet or prompt needs work. If "errors caught" is not zero, write down each one and fix the prompt: "Do not state the deposit unless it is on the sheet." Ask the AI to read the week's replies and flag any two enquirers who received different information about the same property. That is a useful consistency check, though you should confirm anything it flags by opening the messages yourself.

Short lets and the same system

If some of your properties are let by the night or the week, the same structure works with a different fact sheet: check-in times, house rules, parking, how the heating works, local transport, emergency contacts. Guest questions repeat even more than tenant ones, and a message template for arrival and a reminder for departure rules cover the bulk. The tutorial on hotel guest messaging tools compares software built for this when your volume outgrows a phone and a prompt. For a handful of properties, the manual draft-and-approve routine is cheaper and easier to keep correct.

Move-out and the deposit: evidence tables from your notes

The end of a tenancy generates the most disputes. You cannot let a model judge damage from photographs, but it can organise the evidence. Keep a standard inventory for each property with a photo list and condition notes from check-in. At move-out, paste your notes and the check-in inventory text and ask for a comparison table: room, item, condition at check-in, condition now, difference, and your estimated cost with the receipt or quote reference. An illustrative row: "Living room, carpet; check-in: clean, no stains; now: red wine stain about 20 cm; difference: stain; cost: $85 professional clean, quote attached". Rows with no evidence on one side are marked "no evidence", and you leave them out of any deduction. Deductions must be reasonable, evidenced and allow for normal wear, and your local rules on deposit handling and timing apply, so check them before sending anything. Ask the AI to draft the covering message in a neutral tone, and read it against the table before it goes.

The same inventory helps at check-in: a draft welcome message listing how to use the heating, where the stopcock is, bin days and who to contact, assembled from your own notes, should cut the number of questions in the first month.

What goes wrong and how to spot it

• A confident wrong fact. The draft says parking is included on a flat that has none. Keep the fact sheet short and current, and tick the figures in every draft for the first month.
• Different answers to similar people. One enquirer gets an offer of two slots, another gets four. Use a fixed prompt and the same window for all, and review a sample of replies each week.
• Drift into screening by impression. A note in your own file such as "seemed reliable" is a signal you are using something other than the criteria. Replace it with the criterion met or not met.
• Too much personal data in the tool. Keep ID numbers, bank details and full referencing reports out of chats. Use property references and first names in prompts.
• Scams. Fake enquiries and requests to move money are common in rental markets. Never let an AI-drafted message ask for payment before a viewing, and tell tenants plainly that you never will.

Setting it up in a weekend

Saturday morning: write one fact sheet per property and a short criteria document. Saturday afternoon: set up a booking page with your viewing windows and write the reply prompt and the maintenance triage prompt into a project. Sunday: test with ten old enquiries from your inbox and edit the prompts until the drafts need almost no changes. Then use it live on the next vacancy for drafts only, and decide after two vacancies whether any simple replies can be automated.

If you want a comparison point on how other small property businesses handle follow-up, see how to automate buyer and vendor follow-up, and for realistic time expectations read how much time AI can save a small business each week. Also take a look at writing property descriptions with AI for advert wording that stays accurate. And if you would like help turning your own process into a setup, that is what my AI implementation consultation covers.

Landlord questions about AI and tenant messages

Can AI decide which applicant to accept?

It should not. Use it to answer questions, collect the same information from everyone and book viewings. Choosing a tenant should rest on written criteria you set in advance, applied the same way to every applicant, with a person making the decision.

What should my screening questions avoid?

Avoid anything about protected characteristics such as family status, religion, ethnicity, disability or nationality, and anything that steers people towards or away from a property. Ask about move-in date, number of occupants, income evidence, pets and viewing availability, and ask the same of everyone.

Is it fine to let AI send replies automatically?

Start with drafts you approve. Automatic replies suit simple facts such as address, rent, availability and viewing slots. Anything about repairs, deposits, notices, safety or disputes should be read by you before it goes out.

Which personal details can I paste into an AI tool?

As few as possible. Use first names or initials and property references. Do not paste identity documents, bank details or full credit reports. On an individual plan, switch off the setting that lets the vendor use your chats for model training; business plans have it off by default. Check the current terms.

Further reads

• Does a Five-Person Business Really Need an AI Policy? — A short written rule for how you use AI.
• Campervan Rental Software: Handover Checklists and AI Enquiries — Handover checklists and enquiries for a rental fleet.
• AI Tools for Real Estate Brokers: From Transactions to Agent Support — What lettings and sales teams can ask of AI tools.
• AI Policy for a Real Estate Brokerage: What Agents May and May Not Do — Rules for what people may and may not do with AI.
• AI Receptionist vs Front-Desk Hire: The Real Cost for a Salon — When an AI front desk beats hiring, with real costs.
• AI for Real Estate Brokers: Compliance File Review Step by Step — A checklist review idea you can borrow for tenant files.
• Can ChatGPT Draft a House Rental Lease? What Landlords Must Check — Prompts that make ChatGPT flag local-law terms, a check for invented numbers, and the pre-signing list for a house let.
• AI Software for Landlords: Rent, Maintenance and Reminders Compared — TenantCloud, TurboTenant, RentRedi, Innago, Baselane and Buildium compared on price per unit, free tiers and what the AI really does.
• RealPage AI Leasing Agent vs a Standalone Chatbot for Small Portfolios — Knock's AI leasing agent versus a standalone chatbot for a small apartment portfolio: PMS fit, voice, price transparency and exit terms.
• AI Real Estate Investment Analysis: Vet a Rental Deal Step by Step — A rental-deal checklist for small investors: rent comps, assumptions table, cash flow, a seller cap rate check and the AI maths errors to catch.
• AI Tools for Landlords: Messages, Lease Checks and Application Fraud — A small landlord's guide to AI for tenant replies, lease checks and fake application documents, with the legal hard stops and a human decision log.
• AI for Apartment Leasing: A Day Plan for Tours, Follow-Ups, Renewals — An hour-by-hour leasing routine with tour preparation, message templates, renewal checks and human decision points.
• Fake Bank Statements on Rental Applications: A Checklist to Spot AI Fakes — A landlord's checklist for catching AI-made payslips and bank statements: internal consistency, source checks, and why no detector should decide alone.
• Landlord Lead Generation for Letting Agents: AI Outreach, Not Spam — Research local landlords, write personal letters and emails, sort recipients by PECR category and run a short follow-up cadence with a clean opt-out.
• Property Investment AI Tools: A Stack for a Five-Property Portfolio — A five-property investor can start with free deal and bookkeeping tools, then pay only for the workflow that repeatedly saves review time.
• Rent Increase Letters With AI: Notices That Follow the Rules — A free rent increase letter template with merge fields, plus the notice-rule check that stops an AI draft going out with the wrong dates or wording.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: ChatGPT and Claude vendor privacy and pricing pages (checked 4 November 2026). Example figures are illustrative assumptions.