Adel Atya Talk to the demo
Blog · Real Estate

AI receptionist for real estate agents: qualifying buyers at 9pm

Adel Atya · 17 November 2026

An AI receptionist can answer a buyer's 9pm call in about a second, ask your qualifying questions, and book a viewing into your calendar. It should not quote a valuation, give advice, or decide who is a serious buyer.

Below: what it asks, why the first call matters in this trade, what it must never say, and when you don't need one.

What can an AI receptionist do for a real estate agent at 9pm?

It does the part of the call that doesn't need you. The real estate page lists four calls it handles:

The agent asks the questions you wrote. It does not judge the buyer. You read the answers over coffee and decide who gets the first call back.

Why does the first call matter so much in real estate?

The National Association of Realtors 2025 profile (summarised by Virginia Realtors) reports that 88% of buyers purchased through an agent or broker. 43% used an agent they found through a referral and 18% went back to an agent they had used before.

The NAR 2026 highlights page puts the seller figure at 92% using an agent. Referrals and repeat clients are a large share, so the calls you can least afford to drop are the others: a sign call or portal enquiry from a stranger who has not picked an agent yet.

On speed, the best-known research is old and not specific to property. A 2007 study of more than 15,000 web leads found the odds of reaching a lead within five minutes were about 100 times the odds at 30 minutes (summary at Callingly). That summary notes it measured reaching a person, not closed sales.

My own demo shows what fast looks like. On the Blue Ridge Plumbing demo, a missed call gets a text in 8 seconds and an AI callback in 52. A portal enquiry has the same clock running while the buyer compares three agents.

Read those as direction, not a forecast for your market.

What does a 9pm buyer call sound like?

Here is the shape of the call, as a scene and not a client story. The questions are the ones you set, so yours will differ.

A buyer sees your sign at 8:45pm and calls at 9. The agent answers in about a second, says it is an assistant, and asks what they are looking for. A typical script then covers:

  1. Which property or area they called about.
  2. Budget range.
  3. Timeline: this month, this quarter, "just looking".
  4. Whether they are pre-approved or paying cash.
  5. Whether they need to sell a home first.
  6. Which of your open slots suits them for a viewing.

If they ask "what is it worth?" or "will they take less?", the agent says it cannot answer that and offers a call with you. That is a hard rule in the script, because a number quoted on a property nobody has seen can cost you a listing.

Every answer is written to the lead. You wake up to a booked viewing and six facts.

What should an AI receptionist never say?

Write these rules into the script before launch:

My test suite is built on plumbing calls: it runs 38 simulated callers, 114 calls per run, for about $0.25 a run. Its first run caught 5 missed emergency handoffs in 41 plumbing calls on a cheaper model.

Real estate needs its own test calls for the rules above, such as a caller pushing for a price. See how I test AI phone agents.

When is an AI receptionist not worth it for an agent?

Be honest about these cases:

What does it cost, and how long does it take?

A voice agent is $900 to build, or $600 for a founding client, then $450 a month with 500 call minutes included, and 15 cents a minute after that. In year one that is $900 + 12 × $450 = $6,300 at full price. Setup takes five to seven days, including your listings, your qualifying questions and your calendar rules. See the pricing page.

Whether that pays depends on your own numbers: your commission, and how many stranger calls you miss. Count those first.

Checklist before you decide