Can AI answer tenant messages for a landlord, and what are the risks?
Yes, AI can draft and even answer routine tenant questions such as trash day, how to report a repair or how to pay rent, and that saves real time. The risks are a reply that promises something you did not agree to, a reply that breaks fair housing or notice rules, a missed emergency, and texting without consent. Keep a person on anything about money, legal notices, applicants, accommodations or safety.
Sort every message into one of three lanes
The safe way to use AI with tenants is to decide in advance which kinds of message it may answer alone, which it may only draft, and which it must hand to you right away.
| Lane | Typical messages | Who answers |
|---|---|---|
| Routine information | Trash and recycling days, parking rules, office hours, where to send rent, how to submit a repair request | AI can answer from a fact sheet you wrote |
| Needs your decision | A payment plan, a late fee question, a lease change, a repair that costs money, a move-out date | AI drafts, you edit and send |
| Person only | Gas smell, fire, flooding, no heat in winter, a disability accommodation request, a harassment complaint, anything about an applicant or a legal notice | You, immediately, with AI kept out of the decision |
The risks, one at a time
A promise you did not make
A chat that is trying to be helpful may write "we can waive that late fee" or "maintenance will be there tomorrow." Tenants reasonably treat a message from your account as your word. Tell the AI in its instructions never to promise money, dates or exceptions, and to say instead that you will reply.
Fair housing
The federal Fair Housing Act makes it unlawful to make or publish any notice or statement about a dwelling that shows a preference, limitation or discrimination based on race, color, religion, sex, handicap, familial status or national origin. A reply such as "this building is better for professionals than families" is that kind of statement, whoever typed it. The same section covers discrimination because of a disability, which is why a request for an assistance animal or another accommodation belongs in the person-only lane. Read assistance animals and pet policies before you set rules for those requests. Many states and cities add more protected groups, so check yours.
A missed emergency
An AI that answers "thanks, we will look into it" to a message about a gas smell has failed in the worst way. Give it a list of emergency words and an instruction to reply only with your emergency number and to alert you. Then test it with a few made-up messages before tenants ever see it. Your after-hours emergency plan should already say who picks up.
Legal notices
Notices to pay or quit, rent increases and entry notices have state rules on timing, content and delivery. A chat reply is usually not a valid way to give them. Keep notices out of the AI lane entirely; see pay-or-quit notices and rent increase notice rules.
Texting without consent
If the AI sends texts through software, consent and opt-out rules apply. The details are in texting tenants: consent and opt-out rules.
How much time it can save
Example a made-up 30 unit building gets 120 tenant messages a month. 70 are routine, 40 need a decision and 10 are person-only. If a routine reply takes the owner 4 minutes, routine messages cost 280 minutes. If AI answers them and the owner skims each for 1 minute, they cost 70 minutes, saving 210 minutes, or 3.5 hours a month. The 50 other messages still take the same owner time, because that is where judgment is needed.
The saving is real but bounded. AI shortens the easy part of the inbox; it does not remove your job of deciding.
Set it up safely in six steps
- Write a one-page fact sheet per building: address, office hours, emergency number, trash days, parking, how to pay rent, how to request a repair.
- Give the AI written rules: answer only from the fact sheet, never promise money or dates, never discuss applicants, and hand anything unclear to you.
- List emergency and accommodation words (gas, smoke, fire, flood, no heat, disability, service animal, assistance animal) that route a message straight to you.
- Say in the first reply that the tenant is talking with an automated assistant and how to reach a person.
- Keep a log of every AI reply and read a sample each week.
- Start with drafts only for a month before you let it answer anything on its own.
How DoorKeeping approaches it
DoorKeeping is built on the same rule: the AI helps, the owner decides. Payments, signatures, legal notices and applicant decisions wait in your approvals queue. Your AI app can reject an item but can never accept one; accepting happens only on doorkeeping.com when you tap Accept. Every action your AI app takes is written to a log that cannot be changed. Today the DoorKeeping connector reads your portfolio, rent roll and approvals; it does not send messages to tenants. See security for the details.
Sources
Questions
- Can I let an AI chatbot reply to my tenants automatically?
- For routine questions answered from a fact sheet you wrote, yes, with a log you review. Anything about money, legal notices, applicants, accommodations or emergencies should go to a person.
- Am I responsible for what an AI says to my tenant?
- Treat it that way. A message sent from your account reads as your word, and fair housing rules apply to statements whoever writes them. Ask your attorney about your own situation.
- Should I tell tenants they are talking with AI?
- Yes. Saying so in the first reply, with a way to reach a person, sets the right expectations and avoids confusion when a question needs you.