AI Receptionist for Realtors: Cost, Call Routing and Showing Cover

An AI receptionist costs a realtor roughly $30 to $270 a month at list prices, with the bill set by how each vendor counts usage: calls, minutes or distinct callers. Route new buyer and seller enquiries to a message or booked call, send showing requests through pre-approved slots, and send offers, price talks and anything urgent to you immediately.

Cost is rarely what goes wrong. The expensive mistakes are an assistant that promises a showing the seller has not agreed to, answers "is it a safe area?" with an opinion, or texts a caller who never agreed to be texted. All three are routing and wording failures, not pricing ones, so the call map comes before the plan: a plan that books straight into your calendar is only safe once you have decided which slots it may offer.

Follow me on Instagram@sagnikteaches

Plan prices for four services, and what the billing unit means

Figures are USD list prices read on the four vendors' own pages on 1 November 2026. Plans change, so confirm before you buy.

Connect on LinkedInSagnik Bhattacharya

The billing unit matters more for a realtor than for most businesses because calls cluster. A new listing, an open house or a price drop can triple your volume for a week. Per-call and per-minute plans track that spike; per-caller plans absorb repeat callers, which suits a hot listing that the same interested buyer rings three times.

Subscribe on YouTube@sagnikteaches

Rosie's page does not publish an overage rate, so if you pick it, ask the vendor in writing what happens past the allowance before a busy spring.

A solo agent's bill against a three-agent team's

Here are two illustrative cases using the list prices above. Neither is a promise of what you will pay; swap in your own counts.

A solo agent with 40 calls a month

Say 40 calls averaging two and a half minutes, so 100 minutes. On Smith.ai's free plan, 25 calls are free and 15 more at $3 come to $45. On Dialzara Lite, 60 minutes are included and 40 extra at $0.48 add $19.20, so $48.20 in total. Goodcall Starter is $79 flat. Rosie Professional is $49 with 250 minutes, and the 100 minutes sit well inside it. For this agent the range is $45 to $79, and the decision turns on features: if you need calendar booking, neither Rosie's cheaper plan nor Dialzara Lite will do it.

A three-agent team with 150 calls a month

Say 150 calls from about 110 different numbers, averaging two and a half minutes, so 375 minutes. Smith.ai Pro at 150 calls is $270. Dialzara Business Pro costs $99 plus 155 extra minutes at $0.45, which is $168.75; Business Plus at $199 covers everything with 125 minutes spare. Goodcall Starter would cost $79 plus 10 extra callers at 79 cents, so $86.90, or the $129 Growth plan for headroom. Rosie Scale is $149 and includes calendar booking. So the team's range is about $87 to $270, with the lowest price coming from per-caller billing.

Add one more number: your time. Reading summaries and returning calls takes about 10 minutes a day. A typical agent should treat that as the real operating cost, and should not adopt a tool that creates more messages than it saves.

Sort every call into one of six routes

Before you configure anything, label a month of real calls. Most fall into six groups, and each needs a different action. The routing table below is a starting point.

Smith.ai says known spam does not count towards your calls and that up to 10 per cent of calls can be removed as spam. Goodcall does not count spam or wrong numbers as customers. Both help a realtor whose number appears on lawn signs and listing portals.

Showing cover: the weekend problem

Showing requests peak when you are least able to answer, which is the case for an assistant. The goal is to capture enough detail that you can say yes or no in one minute. Here is a filled-in intake, illustrative:

Property: 12 [street] (from your approved list)
Caller: [first name], mobile ending 2290
Wants to view: Saturday between 10 and 1, or Sunday afternoon
Working with an agent: no
Pre-approved or cash: caller says pre-approved, not verified
Group size: two adults, one child
Asked about: parking, whether pets are allowed
Texting consent: yes, caller agreed on the call
Assistant said: "I'll pass this to the agent, who will confirm by text or call. I can't confirm a time yet."

Note what is absent: no lockbox code, no confirmation, no comment on the seller's availability. If the seller needs notice or must approve showings, the assistant offers to take the request and nothing more.

For a team, keep a pre-approved slots calendar per listing. Rosie books directly from its $149 plan and Smith.ai connects to Calendly, so you can expose only the windows the seller has agreed. Block times the seller has refused, and leave a buffer between slots so you are not double-booked.

Fair-housing-safe answers: what it may and may not say

Fair-housing law covers selling and renting. The Fair Housing Act's official overview lists seven protected classes: race, colour, national origin, religion, sex, familial status and disability, and local rules may add others. An assistant speaking for you must not steer callers or comment on who lives somewhere. That is a rule for wording, not just a policy.

• "Is it a safe area?" Say: "I can't give opinions on neighbourhoods. I can share the listing details, and the agent can point you to official sources." Never say "very safe" or "quiet, mostly families".
• "Are there good schools?" Say: "I'd suggest checking the school district's official information; the agent can talk it through."
• "Would my family fit in?" or any question tied to a protected class: do not answer; offer a call with you.
• "Is it good for a wheelchair user?" Give only verified facts, such as "the listing says step-free entrance", and offer to ask the agent for accessibility details. Do not guess.
• Describing the buyer. The summary should record what the caller said they need, not the assistant's impressions of who they are.

The fair-housing checklist for leasing chatbots carries a longer list of risky phrases, and its test approach works for phone assistants too. Have a broker or adviser review your do-not-say list.

Texting and callbacks: where the TCPA enters

An inbound call that the assistant answers is not the main texting-law risk. The risk starts when the assistant, or you, sends a text or places a call afterwards. A regulatory ruling in February 2024 confirmed that the TCPA's restrictions on artificial voices cover AI-generated voices, so outbound AI-voice calls need the right consent, and follow-up texts carry consent rules of their own. The rules differ by message type and place, so take this as a prompt to check with an adviser, not as advice.

1. Ask on the call: "May I text you the listing link? You can reply STOP at any time." Have the assistant record yes or no.
2. Log the consent in the summary with the date, because "he said yes on the phone" is hard to prove later.
3. Honour STOP the first time, across every tool you use. Goodcall can text but cannot receive replies, so a caller who texts STOP back will not be seen by the assistant; route STOP handling through your CRM or messaging number.
4. Keep outbound AI follow-up calls separate. The tutorial on AI voice follow-up and the TCPA covers setup and scripts for that.

Four more situations, scripted

A sign call at 9pm

Caller: I'm outside 12 [street]. Is it still for sale?
Assistant: Thanks for calling [agent name]'s office. I can tell you the listing is on the market as of the information I have, but I can't confirm the price or availability tonight. Can I take your name and number so the agent can reply in the morning with details?

This avoids the stale-listing problem. If your feed updates daily, the assistant is only as current as its last update, which is why it should say "as of the information I have".

An offer deadline

Another agent rings at 6:40pm about an offer expiring at 7. The assistant must not take an offer, discuss terms or say the seller is flexible. Rule: any mention of offer, deadline or counter transfers at once to you, and if you do not answer within 20 seconds it texts you the details marked urgent.

A caller asking for a price reduction

Caller: Will the seller take less?
Assistant: I can't speak for the seller or discuss price. I'll pass your question to the agent, who can respond. What's the best number to reach you?

A rental enquiry on a sales number

Not everyone calls the right business. The assistant should say what you do and do not handle, take a number if you might refer the caller, and end the call quickly. Count these in your first month; if they exceed 10 per cent of calls, change the number on your listings.

Testing before it goes live, and what to measure

Place 20 test calls: five buyers, three sellers, five showing requests, three agents, two neighbourhood-opinion questions and two attempts to get price or access information. Mark each pass or fail. Common failures are the assistant volunteering a schedule it does not know, agreeing the seller is motivated, or saying "family-friendly street" because the listing text did.

Fix the last one at the source. If your listing descriptions contain phrases that would breach the do-not-say list, the assistant will repeat them. Edit the data you feed it, not just the instructions.

Once live, measure four things monthly: calls answered outside working hours, enquiries that became booked showings or valuations, wrong or stale answers reported, and time you spent on summaries. The demo test questions help you compare vendors, and the virtual assistant or AI split helps you decide whether a person should cover part of the load.

Choosing a plan by the kind of agent you are

• Solo agent, mostly buyer calls: start on a free or lowest per-call tier, keep overage under control, and add calendar booking only when weekend requests justify it.
• Listing-heavy agent: per-caller billing absorbs repeat callers on hot listings, but test what happens when volume spikes.
• Team with an ISA or admin: use the assistant as a night and weekend layer, route daytime to the person, and choose a plan with a CRM link so notes land where the team already looks.
• Agent who lives on the phone: consider forwarding only unanswered calls, so the assistant is the fallback and not the greeter.

Whatever you pick, run a two-week trial on forwarded overflow calls first. It lets you compare real calls against the plan limits before you commit, and it protects the relationships built over years from a bad first impression.

Commission at stake when a buyer's call goes unanswered

Take the three-agent team from earlier. Suppose 30 of its 150 monthly calls arrive when nobody can answer: evenings, weekends and while everyone is at a showing. If a third of those callers are genuine new enquiries, that is ten people a month. Say half hang up without leaving a message and ring another agent. That loses five enquiries. If one in ten enquiries becomes a closed sale and the average commission to the team is $9,000, those five missed calls are worth about $4,500 a month in expected income, set against a plan of $87 to $270. The sum is illustrative and every number in it is yours to replace, but it shows why the plan fee is usually the smallest figure in the decision. The figure that deserves attention is the fraction of calls that are genuine enquiries: it comes from your log, not from a vendor brochure.

A weekend setup, step by step

You can have a working assistant in two days if you prepare the material before you open the vendor's dashboard.

1. Friday evening, 45 minutes. Export a month of call logs from your phone provider and label each call by the six routes above. Count them. This decides the billing model.
2. Saturday morning, 60 minutes. Write a one-page fact sheet for each live listing: address, price, bedrooms, bathrooms, parking, open-house times, whether showings need approval. Mark it with the date. The assistant should quote only from these sheets.
3. Saturday midday, 45 minutes. Write the do-not-say list: no price talk, no negotiation, no access codes, no neighbourhood opinions, no comment on who lives nearby, no promises about showings.
4. Saturday afternoon, 60 minutes. Create the account, paste the fact sheets and the rules, connect your calendar and set the notification channel. Choose a text or email alert for urgent items.
5. Sunday morning, 90 minutes. Place the 20 test calls and record failures.
6. Sunday evening, 30 minutes. Fix the top three failures, then forward only your after-hours calls to the assistant for the first week.

That is about five hours of your time, one-off. Most of it is writing material you should have anyway.

A mistake that shows up in week one

An illustrative failure: an agent loads fact sheets for six listings and forgets that one sold on Thursday. On Saturday a caller asks about it and the assistant books a viewing. Nothing is dishonest, but the caller makes a wasted trip and ends the day annoyed with your name on the sign. The fix has two parts. Add a rule that the assistant must say "as of [date on the fact sheet]", and add a Friday ritual of five minutes to remove sold or withdrawn listings from the sheets. A sold listing that still gets calls is a small reminder that an assistant's knowledge is only as current as your last update.

Getting call notes into the system you already use

A summary that stays in the vendor's dashboard gets read less often than one that lands in your CRM or inbox. Check three things in the plan before you pay. First, whether the connector writes a contact record or just sends an email; a record lets you set follow-up reminders. Second, whether the summary fields are fixed or editable, because you want matter type replaced by something like "buyer, seller, showing, agent". Third, how a caller who rings twice is handled; ideally the second call attaches to the same contact. Dialzara lists Zapier and Make connections on every plan but calendar sync only from Business Pro, Smith.ai lists native integrations plus Zapier, and Goodcall connects Google Calendar, Google Sheets and HighLevel natively and reaches other CRMs through Zapier. If your CRM is not on those lists, a Zapier-style step will add its own monthly cost, which belongs in your budget. The CRM comparison for agents shows which systems are worth connecting.

When a person should still answer

An assistant is the wrong first voice for some calls. A seller whose listing has just expired, a past client calling about a problem with a closing, or a buyer who has already written an offer all want you. Build a short list of phone numbers that bypass the assistant and go straight to your mobile, and review that list monthly. Simple rule of thumb: if losing the call costs more than the plan costs in a year, a person should pick up.

Questions agents ask before letting an AI answer their phone

Can an AI receptionist book showings by itself?

Yes, on some plans, but keep the decision with you. Rosie books straight into a calendar only from its $149 plan. A safer setup has the assistant collect the property, the preferred times and the buyer's agent status, then offer slots you have pre-approved. Never let it release lockbox or access details.

Is it legal for the assistant to text callers a listing link?

Often, but it turns on consent. Under the TCPA, texts are generally treated as calls, and a 2024 ruling confirmed that AI-generated voices count as artificial voices on calls. Ask the caller on the call whether you may text them, record the yes in the summary, and include opt-out wording. Check the rules that apply to you with an adviser; this is not legal advice.

What should the AI say when asked about a neighbourhood or schools?

Nothing evaluative. Questions about who lives nearby, how safe it feels or whether it is family-friendly can steer a buyer. The assistant should say it can share the listing facts and point to official school and crime-statistics sources, then offer to book a call with you.

How many calls does a typical agent get?

It varies widely, so count yours. Pull a month of call logs and split them into buyer enquiries, seller enquiries, showing requests, other agents and everything else. A solo agent may see 40 to 80 calls, a three-agent team two or three times that. The count decides which billing model suits you.

Further reads

• Real Estate Chatbot Pricing: What Agents Pay for AI Lead Capture — Prices the website chat side of lead capture.
• Follow Up Boss Pricing: What a Small Team Really Pays — What the CRM that receives your call notes really costs.
• Real Estate AI Chatbot Features: A Checklist Before You Buy — Feature checklist if you add website chat as well.
• How to Automate Buyer and Vendor Follow-Up With AI — Follow-up steps once a call has been logged.
• How to Measure an AI Receptionist's Return in the First 90 Days — How to judge payback in the first 90 days.
• How to Set Up an AI Phone Agent for After-Hours Calls — General set-up steps for after-hours phone cover.
• Best ChatGPT Plan for Realtors: Free, Plus or Business? — Free, Go, Plus or Business for a solo brokerage: what each costs, which ones show ads, and what client data each plan can safely hold.
• Claude or ChatGPT for Realtors: Which Assistant Fits Your Day? — Five realtor tasks to run on both assistants, what each plan costs, how each reads a 90-page disclosure pack, and the privacy settings that matter.
• AI Transaction Coordinator: Which Real Estate Tasks to Automate — Sort transaction paperwork, reminders and chasing into safe AI drafts and human decisions before automating a deal.
• Best Free AI for Realtors: Three Jobs, the Limits and When to Pay — Three jobs a realtor can do on free AI plans, a fair-housing check for every output, and the signs that a paid plan is worth it.
• ChatGPT Training for Real Estate Agents: A One-Week Team Plan — A team leader's one-week plan for training agents on ChatGPT: daily sessions, a rules sheet, five workflows and an adoption check.
• ChatGPT for Realtors: Brokerage Rules for Emails, Disclosures, Follow-Ups — A brokerage owner's rulebook for agent ChatGPT use: what may be summarised, what must be redacted, the fact-check pass and five reusable prompts.
• Open House Sign-In Sheet With AI: Follow-Up Within the Hour — A printable sign-in sheet, consent wording, buyer-type message templates and a 60-minute plan for personal follow-up after every open house.
• Real Estate Sphere of Influence: A Past-Client Touch Plan With AI — A three-tier contact list, a 12-month touch calendar and AI-drafted templates for anniversaries and market notes that keep past clients warm without spam.
• Receptionist Service for Social Media: Can AI Answer DMs and Comments? — What an AI receptionist can reliably do in Instagram and Facebook inboxes for a professional practice, where a person must step in, and how to test it.
• Can Realtors Use HeyGen Avatars for Listings? Rules and Setup — How a realtor can put their own HeyGen avatar on listing updates, what to disclose, and how to avoid MLS, brokerage and fraud trouble.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Smith.ai, Goodcall, Rosie and Dialzara pricing pages (checked 1 November 2026); the official Fair Housing Act overview for the protected classes; the February 2024 regulatory ruling on AI voices under the TCPA; Goodcall's FAQ for integrations and texting.