Skip to main content
RoutMe

Buyer guide

AI chatbot for real estate agencies

By the RoutMe team. Last updated .

A real estate chatbot is most useful at the start of an enquiry. It can answer questions about a listing from your own data, explain how viewings work, and collect what the person is looking for, so an agent picks up a conversation that already has context. It should not negotiate, advise on legal or financial matters, or state anything the listing does not say.

The enquiry problem

A person browsing listings often messages several agencies at once. The first useful reply matters, and so does what happens next. Many enquiries are routine: is it still available, how many bedrooms, are pets allowed, what is included in the rent, can I see it this weekend. A chatbot handles those quickly. The real estate industry page outlines what RoutMe covers.

Listing questions from the source

The assistant needs a trustworthy source for listing facts: price or rent, size, features, status and terms. Where your listing system or website has an API or feed, a read-only lookup returns the current record for one property. Where it does not, listing descriptions added as documents work for stable details, but status such as "under offer" should come from a live source, because a stale "available" is worse than no answer. The distinction between documents and live lookups is covered in RAG vs tools for customer support.

The instruction to the assistant should be strict: describe only what the record says. If asked about the neighbourhood, school catchments or investment potential, it should say it cannot advise and offer to connect the person with an agent.

Collecting requirements

Instead of open conversation, give the assistant a short list to gather:

  1. Whether the person wants to rent or buy.
  2. Area or areas of interest.
  3. Budget range, as the person states it.
  4. Number of bedrooms or other must-haves.
  5. Timing and any constraints.
  6. Preferred contact method and times.

Asking in a consistent order produces comparable notes. A flow can collect the answers and pass them to your CRM or inbox so the agent starts informed. Qualification rules, such as which enquiries go to which negotiator, stay in your own logic and not in the model's discretion.

Viewings

Viewing logistics split into two parts. Process information, such as how long a viewing lasts, what to bring and whether the agent accompanies the visitor, comes from your documents. Actual slots depend on agents' calendars. If you can read availability, the assistant can offer real times, similar to the appointment booking chat pattern. If not, it collects preferred times and passes them to a person to confirm. Be clear with the visitor which one is happening so they do not assume a booking exists.

Where a human takes over

Hand off when the conversation moves to:

  • Offers, price negotiation or discounts.
  • Contract terms, deposits, legal conditions or tenancy law.
  • Mortgage, tax or investment questions.
  • Complaints or disputes.
  • Any request the assistant is unsure about.

Pass the listing, the requirements collected and the transcript. The checklist in when an AI agent should hand off to a human generalizes this. Also decide who owns the inbox in the evening and on weekends, since that is when enquiries often arrive.

Channels

Property searchers frequently use WhatsApp, and a chat widget on listing pages catches visitors who have not shared a number. WhatsApp has its own rules: opt-in, a 24-hour customer service window and approved templates for outbound messages outside it. See WhatsApp vs website chat and the WhatsApp compliance page. Follow your local rules on marketing messages and data protection for prospects as well, and take advice where unsure.

Reviewing quality

Read a sample of conversations each week. Look for places the assistant filled a gap with an assumption, enquiries that stalled before reaching a person, and questions that keep appearing without a good answer. Those point to missing data or documents. How to measure AI support quality shows how to define the numbers without borrowing benchmarks.

How RoutMe approaches this

RoutMe agents read listings and availability through MCP or APIs, answer from your content, and pass qualified enquiries to your team in a shared inbox with the transcript attached. The agent is read-only by default and does not make pricing or legal statements beyond what the listing says. Plans are on the pricing page.

Sources

Explore the product: RoutMe for Shopify, solutions and the documentation.

Frequently asked questions

Does it qualify leads?

It can collect the details you specify and pass them on. The rules for what counts as qualified stay in your flow logic or with your team.

Can it keep listings up to date?

It reads from your listing source rather than storing its own copy of status, so freshness depends on that source. Documents added as snapshots need to be refreshed when they change.

Can it book viewings automatically?

Only if you provide a way to read and write calendars and design that flow deliberately. By default it shows what it can read and passes the request to a person.

Should it talk about prices?

Only as stated in the listing. Negotiation and advice should go to an agent.

Put these ideas to work