An AI agent for real estate is software that handles the follow-up work around your deals. It answers new leads, keeps your CRM up to date, offers showing times, tracks contract-to-close dates and drafts updates for your clients. It works from your inbox, calendar, CRM and deal files. You keep the advice, pricing, negotiation and anything you sign.
Most agents already use AI to write. In the 2026 Technology Report from the National Association of Realtors (NAR), released September 22, 2026, agents who use AI said they use it most for listing descriptions (75%), social media posts (56%) and emails and follow-up (52%). The same survey found that 81% of agents take on new technology mainly to save time, and 63% call the learning curve their biggest hurdle.
A chat tool writes a listing description when you ask for one. The work that slips on a busy team is different. It is the lead who wrote at 9 p.m. and heard nothing, or the inspection date that arrived before anyone noticed. An agent can own that work, because it checks on a schedule instead of waiting for you to ask.
Here is the split we suggest for a small team. Treat it as a starting point. Move a job into the agent's column only after its drafts keep going out unchanged.
| Real estate job | Who does it | What the agent needs |
|---|---|---|
| First reply to a new online lead | The agent, using a reply you wrote and approved once | Your lead inbox or CRM |
| Logging each lead's source, price range, timeline and next step | The agent | Your CRM |
| Follow-ups to leads who went quiet | The agent drafts, you approve | Your CRM and email |
| Offering showing times | The agent checks your calendar and drafts the request, you approve | Your calendar and email |
| Tracking contract-to-close dates | The agent, with a reminder to you before each date | The signed contract and every addendum |
| Weekly updates to sellers on showings and feedback | The agent drafts, you approve | Your CRM, calendar and showing feedback |
| First drafts of listing descriptions and social posts | The agent drafts, you edit | Property details and the photos you choose |
| Monthly check-ins with past clients | The agent drafts a list with a note for each, you approve | Your past-client list |
| A short brief before a listing appointment | The agent | Your notes, past emails and the reports you run |
| Pricing advice, offers, negotiation and contract terms | You | Nothing |
| Answering your phone | Not this kind of agent | A person, or a service built to answer calls |
What can an AI agent do for a real estate agent?
An AI agent can reply to new leads, keep your CRM current, follow up with leads who went quiet, offer showing times from your calendar, track contract deadlines, draft client updates and prepare you for listing appointments. It works on its own schedule, so it spots what is due without waiting for you to remember.
On a small team, each job looks like this:
- New leads. When someone fills out a form on your website or a listing site, the agent sends the first reply you wrote, soon after the lead comes in, at any hour. It logs where the lead came from and what they want, then sets the next step in your CRM.
- Keeping records. NAR's survey found that 46% of agents use a CRM, and a CRM helps only when someone keeps it filled in. After each email, text or showing, the agent logs what happened and the date of the next touch, so nobody has to rebuild the pipeline from memory on Monday.
- Showings. The agent checks your calendar, drafts a request with times you are actually free and keeps track of what got confirmed. If you use a showing service, it can read the confirmations that land in your inbox.
- Contract-to-close dates. The agent reads the dates in the signed contract, such as the inspection period, the appraisal, loan approval and closing. It puts each one on your calendar and reminds you a few days ahead.
- Client updates. Sellers want to know how many people saw the house and what they said. The agent pulls that together once a week into a draft for you to check and send.
- Listing copy and posts. The agent drafts from the property details and the photos you pick. You edit before anything goes live.
- Past clients. Once a month, the agent lists who is due for a check-in, with a reason and a short note for each.
The same pattern works in other industries. Our guide to AI agent use cases for business shows the first job we would hand an agent in eight of them.
Which real estate work should stay with you?
Advice, pricing, offers, negotiation, contract terms and anything you sign should stay with you. So should any message that makes a promise to a client, until you have approved enough of the agent's drafts to trust it with that kind. If your brokerage reviews ads or client emails, anything the agent drafts goes through the same review.
A few more things belong with you:
- Anything about value, such as what a home is worth, what to offer and whether to take an offer. That judgment is what clients hire you for.
- The conversations that build trust, such as the first real call with a serious buyer, the listing appointment or the hard talk after a bad inspection.
- Listing copy before it goes live. Read every line. The agent can get a detail wrong, and the listing carries your name and your license.
- The honest answer. If a lead asks whether they are writing to an AI, the answer is yes. We hold our own agents to that rule.
For a step-by-step way to decide what runs on its own and what waits for you, see our guide to human in the loop for AI agents.
What can go wrong when an AI agent tracks your deals?
The failure to plan for is the missed item: an agent can only track what it can reach. A date inside a document it cannot open never makes it to your calendar. Other risks are leads it files in the wrong place, drafts that sound off and messages that go out with nobody checking them.
We learned this with our own agents. Our proposal agent reads public bids and warns one of our owners about early deadlines. One city bid needed a notice of intent to bid more than a week before the proposal was due. The documents sat behind the city's vendor login, which the agent could not use, so it never saw the date, and the deadline passed. The fix was a rule: when documents sit behind a login the agent cannot use, a person fetches them. The full story is in what our bid agent reads, drafts and hands back.
Real estate has the same trap. Contracts, counteroffers and addenda often live in a transaction platform or an e-signature account. Before you count on an agent for deadlines, make sure it can open every one of those documents, or decide who forwards them. Keep your own check on the dates that cost the most to miss until the agent has caught every date on a few deals.
The other risks have simple checks:
- Wrong labels. A serious buyer filed as "just browsing" gets no follow-up. Look over the agent's labels once a week.
- Drafts that sound off. Approve every message at first, so a stiff or wrong draft never reaches a client.
- Too much access. An agent with a login to your email can send from it. Decide what it may read and what it may change before you connect anything. Our guide on what access an AI agent should have walks through it system by system.
What does an AI agent for real estate cost?
Published prices we read on November 4, 2026 run from $20 a month for ChatGPT Plus, to $58 to $69 per user a month for a real estate CRM with AI features, to $1,995 a month for a full-time real estate virtual assistant. A done-for-you agent from VantaSoft starts at $1,000 to build, plus $200 a month.
| Option | Published price | What it takes off your plate |
|---|---|---|
| ChatGPT Plus | $20 a month | Writing when you ask: listing drafts, emails, posts. It does not watch your inbox or calendar. |
| Follow Up Boss | Grow: $69 per user a month ($58 billed yearly). Pro: $499 a month for 10 users ($416 yearly) | A real estate CRM with automated follow-up, plus AI summaries, suggested tasks and message drafts at no extra charge |
| Structurely | No published plan price ("Talk to Sales"), plus use from $0.02 a text or AI email | AI that works your leads by phone, text and email and books appointments |
| Real Estate Assistant | $1,995 a month | One full-time virtual assistant or inside sales assistant, 40 hours a week |
| Transactly | $399 a transaction with no monthly fee, or $379 a month with one transaction included | A transaction coordinator for each deal |
| VantaSoft | Build from $1,000 one time, then hosting and maintenance from $200 a month | An agent built around your process, your tools and your approval rules, which we keep running |
Each of these covers a different slice of the work, so compare what each one takes off your plate as well as the monthly number. ChatGPT writes when you ask. A CRM's AI works inside that CRM, and some of Follow Up Boss's AI features rely on its calling add-on, which costs extra on the Grow plan. A virtual assistant or a transaction coordinator is a person, with judgment and a phone. An agent works across your inbox, calendar, CRM and files on its own schedule.
For a sense of what agents spend now, NAR's survey found that 36% spend $50 to $250 a month on technology and 22% spend more than $500. Whatever you pick, add your own time approving drafts, which is highest in the first few weeks. For the full breakdown of what goes into having an agent built, see how much an AI agent costs.
Should you get an AI agent or a real estate virtual assistant?
Get a virtual assistant when the work needs a person: phone calls, judgment calls and talking with clients. Get an AI agent when the work runs on rules and a clock: replying, logging, reminding and drafting. Some teams will want both, with the agent doing the routine follow-up so the assistant's hours go to work that needs a person.
A full-time assistant works set hours. At the price above, that is 40 hours a week. An agent can check for new work at any hour, but it cannot pick up the phone or hear that a seller is nervous. Once you write down how a job should be done, the agent follows those notes, and you are not training someone new when an assistant leaves.
For every route to hiring a virtual assistant and the costs that come on top of the rate, read what a virtual assistant really costs.
How do you start using an AI agent in a real estate business?
Start with one job where a miss costs you a deal or a client, such as first replies to new leads or contract deadlines. Write down how you want it done, connect only the systems that job needs, approve everything the agent drafts at first and check its work once a week.
- Pick one job. Choose lead replies if good leads go cold on your team. Choose deadlines if you have had a close call on an inspection or loan date.
- Write your rules. What the first reply says, how many days pass before a follow-up, which hours you take showings and what the agent must never say. We suggest nothing about price or value.
- Connect only what that job needs. Lead replies need your lead inbox or CRM. Deadlines need your signed contracts and addenda.
- Approve at first. Keep a rough count of drafts you send unchanged, drafts you edit and drafts you throw away. Give the agent more freedom only when most drafts go out unchanged.
- Check once a week. Look at which leads got a reply and how fast, which dates made it onto the calendar and which drafts needed edits.
The follow-up rules our own sales agents work by are in our guide to an AI agent for sales. Most of them carry over to buyer and seller leads.
Frequently asked questions
Can an AI agent replace a real estate agent?
No. Buyers and sellers hire an agent for advice, pricing and negotiation, and that work stays with a licensed person. In NAR's 2025 Profile of Home Buyers and Sellers, 88% of buyers bought through an agent or broker. An AI agent takes the follow-up and record keeping around that work, so you have more hours for clients.
Is ChatGPT an AI agent for real estate?
Not on its own. ChatGPT writes a listing description or an email when you ask for one. An AI agent runs on a schedule across your inbox, calendar and CRM, so it notices a lead who went quiet or a date coming up and drafts the next step before you think of it. For more on the difference, see AI agent vs chatbot.
Will my leads know they are talking to an AI?
They should if they ask. The agent sends from your email in your voice, using messages you approved, the same as a draft from an assistant. If anyone asks whether they are dealing with an AI, the honest answer is yes, and that is the rule we give our own agents.
Does an AI agent work with Follow Up Boss or my CRM?
It can if your CRM lets other software connect to it. Follow Up Boss publishes an API, which is a way for other software to read and update your records. Before you buy, ask whoever builds your agent which of your systems it can reach and what it may change in each.
What does it cost to have an AI agent built for a real estate team?
A done-for-you agent from VantaSoft starts at a $1,000 one-time build plus $200 a month for hosting and maintenance. On top of that, plan for your own time approving drafts in the first weeks. For comparison, the full-time real estate assistant in the table above costs $1,995 a month.
Getting an agent built for your team
If leads, showings or deadlines are where things slip on your team, start automating with 30 minutes on your workflows. We build the agent around the CRM, calendar and inbox you already use, set the approval rules with you and keep it running. To see the follow-up role written out, read what our AI sales agent owns and where it stops. For other places to start, see AI agents for business.
