How agencies and property managers can use AI chat on web, WhatsApp and Messenger to answer enquiries quickly, qualify leads and hand off to agents.
Property enquiries are perishable. A buyer who messages three agencies on a Saturday evening usually talks to whoever answers first, and most agents are out at viewings when the messages arrive. This article shows how to use AI chat to answer property enquiries quickly, collect the details a negotiator needs, and pass hot leads to a human at the right moment.
Why enquiry response is a real estate problem
Real estate has an unusual enquiry pattern. Messages come from portals, your website, social ads, WhatsApp links on signboards and Messenger. They arrive at night, at weekends and while your team is in the field.
Three things go wrong again and again:
- Slow first reply. The enquirer has moved on by the time someone responds.
- Repeated questions. Agents answer the same questions about price, availability, pets, deposits, parking and viewing times many times a day.
- Thin lead records. A message that says "Is this still available?" tells you nothing about budget, timeline or financing.
AI chat does not replace your negotiators. It handles the first minutes of every conversation, so the human who follows up starts with context.
What an AI agent should and should not do
The safest approach is to give the AI a narrow, well-defined job.
Good jobs for AI
- Answer factual questions from your own listing information, fee schedules and process guides.
- Ask qualifying questions in a natural order.
- Offer viewing slots and capture contact details.
- Reply in the enquirer's language.
- Route urgent or sensitive conversations to a person.
Jobs to keep with humans
- Price negotiation and offers.
- Legal, tax or mortgage advice.
- Anything where the knowledge base has no answer.
- Complaints from existing tenants or owners.
Choosing your channels
Meet enquirers where they already are. For an international audience, the everyday set is usually:
| Channel | Typical use in real estate |
|---|---|
| Website chat | Listing pages, developer sites, rental search |
| Signboard QR codes, ad click-to-chat, follow-up with prospects | |
| Messenger | Facebook page and ad enquiries |
| Longer questions, document requests |
Keep one knowledge base behind all of them, so the answer about a deposit is the same on every channel.
Build the knowledge base your agent will rely on
An AI agent is only as good as the material it can read. For real estate, prepare:
- Listing summaries. Address, price or rent, size, floor, fees, available date, pet and parking rules.
- Process guides. How viewings work, what documents tenants need, how long checks take.
- Fee and cost explanations. Agency fees, deposits, management fees, in plain language.
- Area information. Transport, schools, local services, as you would give to a client.
- Policies. Opening hours, languages spoken, how complaints are handled.
Keep listings current. An agent that quotes a rent that changed last week creates a worse experience than a slow reply. Assign one person to update the sources whenever stock changes.
Qualify without interrogating
A good chat conversation feels like a helpful front desk, not a form. Ask one question at a time, and ask for details only when they make sense in context.
For a rental enquiry, a sensible order is:
- Which property or area are they interested in?
- Move-in date.
- Budget range.
- Number of occupants and any pets.
- Name and preferred contact method.
For a purchase enquiry, add whether they need financing, whether they have a property to sell, and whether they are buying to live in or to let.
- 01Enquiry arrives
- 02AI answers and asks questions
- 03Lead details saved
- 04Agent on shift takes over
Example of the value
Here is a hypothetical. If an agency receives 300 chat enquiries in a month and a negotiator spends about two minutes on each simply gathering basics, that is 600 minutes, or 10 hours, of repetitive work. If AI handles that first stage, those hours go to viewings and follow-up instead. These numbers are only an illustration; your own volumes will differ.
Hand off to a human at the right moment
Define clear triggers for passing a conversation to a person:
- The enquirer asks to book a viewing or speak to an agent.
- The enquirer mentions an offer, a deadline or a competing property.
- The question is outside the knowledge base.
- The tone suggests frustration.
The handoff itself matters. The customer should know what happens next, and the agent should receive the conversation history so nobody has to ask the same questions again. Be careful with promises: never tell a customer "someone will reply in five minutes" unless your rota can really deliver that.
Keep it accurate, private and compliant
Real estate chats involve names, phone numbers, incomes and sometimes identity documents. A few principles help:
- Ask only for what you need at the first stage. Leave identity and financial documents for a secure human-led process.
- Tell people they are talking to an AI assistant.
- Review conversations regularly and correct wrong answers so the agent improves.
- Respect opt-outs. If you send follow-up messages, only contact people who have messaged you first and give them an easy way to stop.
- Check local rules on marketing consent and data handling with your own advisers.
Doing this in AI SNS CHAT
Here is how the approach above maps to AI SNS CHAT.
1. Create the agent. Use the guided wizard, then set the persona and guardrails, including the "do not" list. You can start from one of the starter personas and adjust it. You can test answers in Playground before going live, and restore points let you roll back if an edit makes things worse.
2. Add knowledge. Under Data sources, add Files (brochures, fee guides), Text snippets, your Website, and Q&A pairs for common questions. If your listings or documents live elsewhere, connectors cover tools such as Google Drive, OneDrive, Dropbox, Box, SharePoint and Notion. In Japan, kintone is supported for read and write.
3. Deploy on channels. From Channels, add the Chat widget to your site, then connect WhatsApp and Messenger. The welcome message is auto-translated into every language the widget speaks. A custom domain (Agent → Settings → Custom domains, Hobby or higher) keeps the widget on your own subdomain.
4. Capture leads. Use the Collect leads Action or a custom form. Leads appear in Activity → Leads, and every lead also creates a contact. In Custom attributes, the description of each field tells the agent when to ask for it, which is how you encode "ask for budget after move-in date". Contacts can be exported as CSV. Integrations need Hobby or higher, and the number of Actions per agent depends on the plan.
5. Hand off. Under Helpdesk → Scheduling (Standard or higher), set shifts per person and timezone; handoffs are auto-assigned to whoever is on shift. Alerts can go to Slack, Teams, LINE WORKS, WhatsApp, email or a webhook. When a chat is handed off, the customer message includes a "what happens next" line based on the rota, and the system never invents a wait time. Staff can take over a chat, and the bot pauses.
6. Improve. Review Activity → Chat logs, rate answers with thumbs up or down, and edit a reply into the ideal answer to train the agent. Analytics show topics and question history, with advanced analytics on Pro.
7. Follow up. Outbound Beta (Standard or higher) can message people who contacted you first, one to one, with a STOP opt-out on every message. WhatsApp sends are limited to the 24-hour window. There are no SMS or email marketing campaigns.
Each reply uses one credit. See pricing, features and channel connections for current details.
Conclusion
Fast, accurate first replies win real estate enquiries. Give the AI a narrow job, feed it current listing and process information, qualify one question at a time, and hand off to a person when the conversation becomes a deal. Measure the result by reading real chats and improving the answers each week.
Frequently asked questions
Can the AI tell a buyer that a property is still available?
Only if that information is in the knowledge base you keep up to date. Otherwise, set the persona to hand off or say it will check with an agent. Keeping listing sources current is the most important maintenance task.
Does AI SNS CHAT work on WhatsApp and Messenger?
Yes, both are available in the Channels list alongside the website chat widget, and they share the same agent and knowledge base. Check the connections page for setup requirements.
What happens when a customer wants a human?
You can set the agent to hand off, and the chat is assigned to whoever is on shift in Helpdesk → Scheduling, which needs Standard or higher. Your team gets an alert and can take over, which pauses the bot.
Where do captured leads go?
They appear in Activity → Leads, and each one also creates a contact that you can export as CSV. If you use a CRM such as HubSpot, Pipedrive or Zoho CRM, read-only lookup tools are available, so check the features page for what fits your setup.


