Conversational Home Search Opens Room for Listing Checks and Viewing Bookings

Home search has moved from search to action

CoStar Group announced on September 8, 2026, that people could use the US rental-housing service Apartments.com inside ChatGPT. Users can describe requirements in a sentence, including budget, number of bedrooms, whether pets are allowed, a balcony, or electric-vehicle charging, and search for matching homes.

Results include available units, current rent, floor plans, required and recurring fees, photos, videos, and 3D walkthroughs. After choosing a home, users can message the property manager, request a viewing, and book a time without leaving ChatGPT.

CoStar said that, after introducing the feature in early August 2026, visits arriving at Apartments.com from ChatGPT increased by about 40%. If there had been 100 visits before, that would be roughly 140, but this is a company-internal figure. The company did not disclose how much actual viewing bookings or lease contracts increased.

In an Apartments.com survey of about 27,000 US adults, 4% said they use or plan to use conversational AI for rental search. Direct use is still limited, but habits may be forming first: 30% said they used services like these daily or weekly, and 41% said they might rely on recommendations when choosing a home.

A longer conversation does not automatically make recommendations neutral. Apartments.com states that the ordering of results in conversations can reflect advertising products. Users therefore need tools that clearly separate why a property was recommended from whether it is an advertisement.

Verification and appointments come before recommendations

Consider Mr. Kim, a licensed real-estate agent working with two staff members near a university district in Seoul. When a student contacts him after seeing a listing advertisement, he asks again about the preferred move-in date, deposit, monthly rent, maintenance fee, and whether pets are allowed.

He then copies information from the listing screen into notes or a chat room and calls the landlord to check whether the property is still available. If it is, he coordinates the schedules of the agent, current resident, and student in sequence, then proposes a time again.

In this process, the same conditions are copied several times, listings with changed prices are discovered late, and evening enquiries roll over to the next day. If a viewing is cancelled the day before, someone must find another time or a similar listing and contact the student again.

With conversational home search, a student can state their conditions once. The service can show what the listing says alongside the time it was last verified, and compare scattered costs such as the deposit, monthly rent, maintenance fee, and parking fee on one screen.

When the student chooses a listing, the service sends set questions to the landlord or responsible agent. Once availability and viewing times come back, it shows the student the options, adds the selected appointment to a calendar, and sends reminders the day before and on the day itself.

Put another way, as conversation becomes more natural, the important thing is not sounding human. If a service cannot verify that a listing is still available today, that its price has not changed, or that the person in charge approved the appointment, a long conversation can create false confidence.

The work left to people is also clear. Agents still need to inspect the condition of the property and its legal status, coordinate terms with landlords, and explain the contract. Rather than building a large service that recommends listings nationwide from the start, a smaller tool for verification requests, appointment confirmation, and change alerts is more realistic.

Elsewhere, products bundle booking with verification

Zillow and EliseAI

US company Zillow added EliseAI’s AI leasing-assistant capability to listing pages for rental operators managing more than 450 homes. The feature supports answers to questions, tour bookings, follow-up guidance, and rental applications.

In a joint analysis covering October 2025 through April 2026, the two companies said people who used the feature were 19% more likely to book a tour, 43% more likely to submit an application, and 24% more likely to sign a lease than people who did not. These figures come from a joint supplier announcement, so they cannot establish that the same results will occur in other markets.

DashQ and Osgood Properties

Canadian company DashQ combined text-message, email, and phone records for Osgood Properties, which manages about 5,000 homes, in one place. It added available viewing times, booking confirmations, rescheduling, and reminders so that enquiries would not remain in individual staff notes.

The company said that 34% of 12,298 enquiries in the first half of 2024 led to viewings, and that it reduced booking work by 559 hours each quarter. That is enough time for one staff member to spend more than three months handling bookings alone, but it is the company’s own count based on a customer case.

HousingAnywhere

Netherlands-based HousingAnywhere verifies landlords’ identities and bank accounts for international students and workers looking for a home from abroad. It holds the first month’s rent and sends it to the landlord 48 hours after move-in, while offering an investigation and refund process when the listing and home differ substantially.

Users pay a one-time protection fee equal to 25–40% of the first month’s rent. This is an example of why users may pay only when a service bundles more than a recommendation badge: identity checks, payment protection, and a process for raising problems after move-in.

Four things you can build now

1. A listing-check inbox with a visible last-verification time

• What it does: It asks landlords or agents, in a set message, whether a property remains available and what the current price is, then records the time of the reply.
• Who uses it: Licensed real-estate agents near university districts who post the same studio listing on several portals.
• Why now: Even if AI recommends properties well, choosing outdated advertisements loses both enquiries and trust.
• First screen: Start with listings that have gone the longest without verification, and show available, awaiting reply, and contract completed states with both colour and text.

2. A real housing-cost comparison receipt

• What it does: It puts the deposit, monthly rent, maintenance fee, parking fee, internet cost, and missing information for each listing onto one page.
• Who uses it: International workers or students who are unfamiliar with Korean rental advertisements and their cost categories.
• Why now: Conversational search expands the candidate list quickly, but final comparison remains difficult because each advertisement presents costs differently.
• First screen: Show three lines first: the monthly amount, the amount needed upfront, and items that remain unverified.

3. An enquiry assistant that collects only unanswered questions

• What it does: It sends registered answers to repeated questions about parking, pets, and move-in dates, then passes only uncertain questions to the person responsible.
• Who uses it: Small officetel management offices that miss enquiries in the evening and at weekends.
• Why now: It has become more important to prevent wrong answers and identify the moments that need a person than to produce natural-sounding answers.
• First screen: Show unanswered questions, waiting time, and the responsible person who needs to verify them, in that order.

4. A viewing-booking and follow-up manager

• What it does: It collects available times from the agent and resident, confirms a viewing, and handles changes, cancellations, and reminders the day before.
• Who uses it: Neighbourhood real-estate offices where studio viewings for students and early-career workers cluster on Saturdays.
• Why now: Home recommendations are becoming faster in domestic services as well, but an actual viewing still ends only after several people confirm their schedules.
• First screen: Arrange today’s viewings, still-unconfirmed appointments, and time made vacant by cancellations in chronological order.

What to verify in 30 minutes today

Call three licensed real-estate agents near a university district and ask how many times they rewrote the same information in the past week because of listing-status checks and viewing coordination. If two of the three say this happens every day, and even one agrees to a one-week trial, it is worth building a screen for “last verification time and viewing appointments” before adding recommendation features.

Why this matters where you are

Check whether home-search services in your market can verify current availability, price, and appointment approval rather than only recommend listings. The rental terms, fees, and people involved in a viewing may differ from the Seoul examples, but the handoff from a suggested property to a confirmed appointment still needs clear records. A small workflow tool can begin with the repeat questions and scheduling steps local agents already handle.