What a real estate AI voice agent does
A real estate AI voice agent answers the calls that come from sign riders, listing portals, your website and your main line, qualifies the caller in a short conversation, answers approved listing facts, and books a showing or a consultation into your calendar while the caller is still on the phone. It then writes the lead, the transcript and the booking into your CRM and alerts the agent on duty. Solo agents, teams and brokerages use it to answer the after-hours and drive-time calls that would otherwise reach voicemail.
This guide covers what the agent should ask, which listing questions it can safely answer, what a month of calls costs on a flat plan, when an outbound reactivation campaign makes sense, and how to decide between an AI agent, a human answering service and an inside sales agent. It is written around the real estate voice agent built on Botcadence, so we are not neutral; the questions and the math apply to any vendor.
Why speed-to-lead decides who gets the showing
A buyer who calls from the curb outside a listing is comparing houses, not agents. If the call reaches voicemail, the next call goes to the number on the next sign or the next portal listing. The common term for this is speed-to-lead: the gap between the first contact and the first real conversation. You do not need a published statistic to see it in your own CRM. Pull last quarter's leads, note the minutes between inquiry and first conversation, and compare conversion for the leads answered within five minutes against the rest.
The calls that get missed are predictable: evenings, weekends, open-house hours when the agent is on site, and any time the agent is showing property. A voice agent is designed for exactly those windows. The inbound voice agent can answer every call at the same time without a queue, follow the same qualification script on each one, and hand hot leads to a person with the transcript attached.
What the agent should ask on the first call
Keep the script to what changes the next action. For most teams that is:
| Question | Why it matters | Where it lands |
|---|
| Buying, selling, or both? | Routes to the right agent and script | CRM lead type |
| Price band or budget | Filters inventory and sets priority | CRM custom field |
| Timeline (now, 1-3 months, later) | Decides whether to book a showing or a consultation | CRM stage |
| Financing status (pre-approved, cash, not started) | Flags leads that need a lender introduction | CRM field, lender referral task |
| Areas or zip codes of interest | Matches to the agent who covers that area | Routing rule |
| Working with another agent? | Avoids wasted showings and ethics problems | CRM flag |
| Best callback number and preferred contact channel | Enables SMS or WhatsApp confirmation | Contact record |
Everything the agent captures should be a structured field, not a free-text note, so your lead scoring rules and your routing can act on it without a person re-reading the transcript. The general method is covered in the lead qualification automation guide, and the lead qualification voice agent page shows the rubric in use.
Which listing questions the agent can answer
The agent answers from the content you give it. For listing questions, that means a property sheet or a feed you approve: beds, baths, square footage, HOA fees and rules, showing windows, parking, school district as listed, and whether the property is still active. Keep the sheet current, because the agent will quote whatever it was last given.
Anything that amounts to advice stays with a licensed person: pricing strategy, offer tactics, what the seller might accept, fair-housing-sensitive questions about neighborhoods, legal or tax questions, and anything about a property not in the approved data. The agent's rule for those is to say so and offer a callback or a transfer. Write those refusals explicitly before launch; they are the most common source of complaints in early deployments.
Booking showings during the call
Booking is where the lead turns into a meeting. Connect the agent to the calendar your team actually uses, through Google Calendar, Calendly or Cal.com, and give it rules: showing lengths, buffer time for travel, which agent covers which area, and which slots are reserved. A qualified buyer hears two or three open times, picks one, and receives a confirmation by SMS or WhatsApp. The agent on duty gets the booking, the transcript and the qualification fields in HubSpot, Salesforce or, through Zapier, the real estate CRM you already run.
For high-intent callers, a warm transfer is often better than a booking. Set a rule such as "pre-approved buyer, timeline now, agent on duty available" and let the agent transfer the call with context instead of scheduling for tomorrow. The human handoff page describes what the receiving agent sees.
What a month of calls costs
Botcadence prices the Voice AI plan at $249 a month with 1,500 minutes included and $25 for each extra pack of 200 minutes. Managed voice setup takes about 5 business days, and you keep your existing numbers and forward them after hours, on overflow or all day. The example below uses stated assumptions; swap in your own call log.
Say a six-agent team receives 300 inbound calls a month that currently reach voicemail or a shared cell phone, averaging 4 minutes each with qualification and booking. That is 1,200 minutes, inside the included 1,500, so the month costs $249.
Say the team adds a reactivation campaign through the outbound voice agent: 1,000 leads from last year who went quiet, of whom 300 answer and talk for 2 minutes on average. That adds 600 minutes, for 1,800 in total: $249 plus two $25 packs, or $299 for the month.
Say the two programs together book 20 showings or consultations. The cost per booked meeting is about $15. Whether that is good depends on your average commission and your conversion from showing to closing, which you know and we do not.
| Program (assumptions) | Calls | Minutes | Plan math | Cost |
|---|
| Inbound only | 300 at 4 min | 1,200 | Within included minutes | $249 |
| Inbound plus reactivation | 300 at 4 min, 300 at 2 min | 1,800 | $249 plus 2 packs | $299 |
| Busy spring month | 500 at 4 min, 300 at 2 min | 2,600 | $249 plus 6 packs | $399 |
Check the pricing page for how ring time and unanswered outbound attempts are counted before modeling a large reactivation list.
Outbound reactivation and consent
Calling old leads is where most of the compliance risk sits. In the United States, automated calls to mobile numbers need prior express consent under the TCPA, do-not-call lists apply, and several states regulate recording. Only load contacts whose consent you can show, call inside permitted hours for their time zone, let the agent identify itself as automated, and honor an opt-out on the spot. The lead reactivation voice agent page covers the campaign setup; treat this paragraph as a checklist, not legal advice.
AI voice agent vs answering service vs inside sales agent
We build Botcadence, so read this section with that in mind.
- Choose an AI voice agent when missed calls cluster after hours and during showings, when you want booking rather than message-taking, and when your CRM or calendar has an integration or a webhook path.
- Choose a human answering service when you want a person on every call regardless of cost, your volume is small and steady, and a message with a callback is acceptable.
- Choose an inside sales agent (ISA) when you have enough lead volume to keep a person busy full time and the conversations need negotiation or local judgment on most calls. Many teams run an ISA by day and the AI agent after hours and on overflow.
The AI receptionist vs virtual receptionist vs answering service comparison goes deeper on cost models.
Multilingual callers
Markets with many Spanish-speaking or other non-English buyers should test language coverage before launch. On Botcadence, chat agents respond across 50+ languages with automatic detection; voice language coverage depends on the voice model you select, so run test calls in each language you need and listen to the results. See multilingual support.
What to measure and common mistakes
Measure weekly for the first month:
- Answer rate on forwarded lines, before and after.
- Qualified leads per 100 calls, using the fields above.
- Showings and consultations booked per 100 calls, and the show rate.
- Transfer quality: did the receiving agent have the context, and did the caller repeat anything?
- Time from call end to CRM record, which should be seconds, not hours.
- Refusal rate on listing questions, which tells you what to add to the property sheet.
Mistakes to avoid: quoting listing facts from a stale sheet, letting the agent speculate about pricing, booking into a calendar nobody checks, forwarding every call to the AI when the office is open and staffed, and skipping the after-hours test calls from a real mobile phone before cutover.
Rollout plan
Start with inbound calls on one line, usually the after-hours forward of the main number to the inbound voice channel, with the CRM and the calendar connected. Add listing data for the properties that generate the most calls. Add reactivation once consent records are clean. Review after-hours bookings weekly before extending to every sign rider and portal number. The real estate industry page shows how the same agent covers website chat and WhatsApp.
If you want to see what the agent does with your own call types, start on the AI voice agents page and run the five-call test: a curbside buyer, a seller inquiry, a question about HOA fees, an out-of-area caller and a caller who asks for a person.