A listing goes live on the property portals and the inbox fills up: "Still available?", "Can I view on Saturday?", "Is there parking?". The agent is out at a viewing, replies in the evening, and the most motivated buyer has already seen something else. In this business, response time matters as much as the property itself.
An AI agent doesn't show homes and doesn't negotiate. But it handles what happens before and after a viewing very well. For what an AI agent is in general, see my AI agent use cases for business. Here, we stay inside the agency.
In short. Five uses: qualify inbound inquiries, offer viewing slots, follow up with buyers after viewings, answer missed calls with a voice agent, draft listings and seller reports. In every case, the human agent keeps control of price, negotiation and choosing between applicants.
Why would a real estate agency use an AI agent?
Because most inbound messages look alike and arrive at all hours. The questions are about information that already exists in the property file (size, fees, floor, availability), and the next step is nearly always the same: check the buyer's plans, then offer a viewing. It's a repetitive task with clear rules, and it has a real cost when it waits.
What are 5 uses of an AI agent in a real estate agency?
1. Qualify inbound inquiries
The agent answers messages from portals, the website and social media using the property file, then asks what your agent would ask: buying or renting, budget, financing (pre-approval or not), must-haves, timeline. It creates or updates the contact record in the agency's software.
Limit: it only answers with what the property file says. A question the file doesn't cover ("has the roof repair been approved?") goes to the human agent.
2. Offer viewing slots
Once the inquiry is qualified, the agent offers open slots from the human agent's calendar, books the one chosen, and sends a confirmation with the address and a reminder the day before.
Limit: group viewings, occupied homes and seller constraints have to be entered in the calendar. The agent can't guess them.
3. Follow up with buyers after a viewing
The day after a viewing, the agent sends a short message asking for the buyer's opinion, offers a second viewing or a call, and files the answer. It also drafts the summary of feedback for the seller report.
FromYour agent
SubjectYour viewing of the apartment on Lilac Street
Day 1
Hi Mr. Garner, thanks for coming to the viewing yesterday. What did you think? If you're interested, I can set up a second viewing or a call with Claire to talk about next steps. If it wasn't right, a word on what was missing will help me find you something better.
Limit: an offer, a price question or a complaint goes straight to the human agent.
4. Pick up missed calls with a voice agent
After hours, or when the whole team is out, a voice agent answers, introduces itself as the agency's AI assistant, notes the request (which property, what plans, callback wanted) and sends a summary to the human agent. It can also offer a callback slot. I explain how phone agents work in my guide to the AI phone answering agent.
Limit: the voice agent handles inbound calls. Calling people who didn't ask to be contacted is cold calling, which is regulated in most countries (in France, by the Consumer Code). Check your local rules before any outbound campaign, automated or not.
5. Draft listings and seller reports
From the property data sheet, the agent's notes and the photos, the AI writes a first draft of the listing, and each month a draft seller report (inquiries, viewings, feedback). The human agent reviews, edits and publishes.
Limit: a listing commits the agency. Mandatory disclosures, floor areas and inspection reports are checked by a person, every time.
What does a qualification workflow look like, step by step?
Here is the typical setup I build for the first use, the one that pays off fastest. Each step says what you should see on screen when everything works.
- 01
Inbound
A message comes in from a portal or the website form. In the automation tool, a new run appears with the message text and the property reference.
- 02
Reading the property file
The agent fetches the property file from the agency's software. The run shows the fields read (price, size, fees, availability).
- 03
Reply and questions
The agent answers the question and adds two or three qualifying questions. The reply goes out from the agency's address, stating that it was written by an AI assistant.
- 04
Contact record
The buyer's answers fill in the contact record: budget, financing, criteria. In the software, the record shows the status "qualified" or "call back".
- 05
Handover
The human agent gets a five-line summary (WhatsApp, email or task) and a link to the record. They decide what happens next.
What does the AI agent do, and what stays with your team?
| Task | AI agent | Human agent |
|---|---|---|
| Answer questions about the property | Yes, from the property file only | Anything the file doesn't say |
| Qualify the buyer | Asks the questions, fills in the record | Judges how serious the project is |
| Schedule viewings | Offers and books open slots | Sets the calendar rules |
| Follow up after viewings | Sends, collects and files feedback | Handles offers and objections |
| Value or discuss the price | Never | Always |
| Choose a tenant or buyer | Never | On objective criteria |
What are the limits of an AI agent in real estate?
- It can be wrong about a property. If it reads an incomplete or outdated file, it answers wrongly and confidently. The quality of your property files becomes a priority.
- Fair housing and non-discrimination. Rejecting a rental or purchase applicant on a discriminatory ground is illegal in France and in many other countries, whether a person or a tool makes the decision. The agent doesn't screen applicants: it collects objective information (budget, financing, timeline) and a person decides.
- Telling people. In the EU, the AI Act requires telling people they are interacting with AI, unless it's obvious. The text published in 2024 says this obligation applies from August 2, 2026. One sentence at the start of a message or call is enough.
- Personal data. Family situation, income, financing: that is personal data, sometimes sensitive. Under the GDPR, only ask for what the decision needs, use tools that don't reuse data and set a retention period. Outside the EU, check your local privacy law.
- Portal terms. Check what your contracts with the property portals allow when it comes to automated replies to the leads they send you.
- The relationship. A seller who trusts you with their home wants to talk to a person. The agent drafts the report; it doesn't replace the call.
Where should a real estate agency start with AI?
- Measure your current response time to portal inquiries over two weeks. That number will tell you whether the agent is worth it.
- Update your property files: the agent will never be better than they are.
- Start with qualifying inbound inquiries, with drafts approved by a human agent for the first few weeks.
- Then add viewing scheduling, then post-viewing follow-up.
- Keep the voice agent for later, once the written flow works.
This is the kind of project I build in my custom AI agents work, often with some automation to connect the agency's software, the calendar and email. We start with a 30-minute call, then a free mockup built on your agency.
Sources checked on October 17, 2026
- EUR-Lex, Regulation (EU) 2024/1689 on artificial intelligence, articles 50 and 113 (published text; check the consolidated version)
- CNIL (French data protection authority), Q&A on using a generative AI system (July 18, 2024, in French)
Frequently asked questions
Can an AI agent replace a real estate agent?
No. It takes over repetitive exchanges before and after the viewing: answering questions, qualifying, booking, following up. Viewings, negotiation, advising the seller and choosing between applicants stay with people.
What is an AI voice agent for real estate?
An assistant that answers the agency's phone when nobody is available, introduces itself as AI, notes the request and sends a summary to the human agent. It handles inbound calls; it isn't meant for cold calling.
Can an AI agent value a property?
I advise against it. A valuation commits the agency and relies on local market knowledge and a visit. The agent can prepare the data (comparable sales, features), but the valuation is done and signed by a professional.
Do buyers need to know they are talking to AI?
In the EU, yes. The AI Act requires telling people they are interacting with AI, unless it's obvious from the context. A clear notice at the start of a message or call is enough.
Which use should a real estate agency start with?
Qualifying inbound inquiries from portals and the website. That is where slow replies cost the most leads, and the flow is easy to check with one human agent.