AI Tools for Real Estate Agents: An Honest Guide

Search “ai tools for real estate agents” and you get page after page of ranked tool lists. None says what a tool costs after the trial, or that a listing description you did not write is still your ad under fair housing law.

This guide is organized by the job. Neither platform built for real estate publishes a price, so you cannot compare an all-in-one against three tools without a sales call. Start where the price is public.

Key takeaway: ChatGPT, Follow Up Boss, Canva and BoxBrownie publish real prices. Lofty and Ylopo make you book a demo. Start with the copy job on a $20 ChatGPT subscription, and remember you stay the advertiser under fair housing law.

The three jobs worth handing to AI in real estate

The job keeps you out of the office, which is where marketing gets done. The Bureau of Labor Statistics says agents “spend much of their time away from their desks to show properties … and meet with current or prospective clients.” Three jobs stall when you get busy. In buying order:

1. Listing and property copy, which a general assistant handles.
2. Lead response and follow-up, needing a CRM with AI or a voice agent billed by the minute.
3. Keeping marketing running while you are out, needing design and scheduling software.

Copy comes first because it is cheapest to test and stalls first. NAR is a trade association surveying its own members, and its 2025 Technology Survey drew 1,241 usable responses from 49,233 invitations, a 2.5 percent response rate, so read its numbers as signals. In it, 41 percent currently use AI.

Tools for listings and property copy

No real-estate-specific listing-copy tool could be priced here, so none is named. Agents reach for the general one, and NAR’s survey put ChatGPT on top at 58 percent.

OpenAI, which sells the subscription, lists ChatGPT Free at $0 a month, Go at $8, Plus at $20 and Pro at “From $100 / month”, all US list prices that vary by region. Go “may include ads.” Paste client information in and it trains the models unless you opt out, and the personal plans carry none of the account controls a brokerage IT policy asks for.

Fair housing exposure is highest here, because the phrases that cause trouble read as ordinary marketing. HUD’s own example is a rental ad carrying the disclaimers “no criminal records!” and “good credit only!”, which HUD says “could deter or discourage applicants on the basis of race” when aimed at particular neighborhoods. A model writes screening language like that without hesitating. Treat every output as a first draft, the same discipline as keeping the drafts in your voice.

Tools for lead response and follow-up

Either a CRM (customer relationship management software, the database holding your leads) with AI bundled in, or a metered voice agent, charging per minute of talking.

Follow Up Boss, a real-estate CRM vendor, is the rare one publishing list prices: $69 per user per month plus tax on its Grow plan. The catch, in their words: “many of the features rely on data produced by the FUB Calling, which is an added charge on the Grow plan.” Calling is $39 per user, so the AI as advertised costs $108 per user per month, by arithmetic on their own two numbers.

Retell AI sells voice agents to developers rather than to agents, so treat its published rates as the reference price for a minute of AI calling: $0.07 to $0.31, plus $2 a month for their number. Asked whether it bills during silence or hold time, the page answers “Yes.” Convert it first: 200 minutes a month, ten a day, is $14 to $62 plus the $2.

One caution about the statistic vendors quote. The twenty-one-times claim about qualifying leads in five minutes versus thirty minutes comes from a document titled “THE INSIDESALES.COM /MIT LEAD RESPONSE MANAGEMENT STUDY”, presented by that vendor’s own chief executive, built on its own CRM data, six companies, copyright 2007, fieldwork in mortgage and insurance instead of brokerage. Its authors wrote “This study did not address close ratios.” Fast follow-up is still good practice, and that deck is not the evidence for it.

Tools for keeping marketing running week to week

Social media was the top lead-generating technology in NAR’s survey at 39 percent, and that queue has to keep running. Canva sells design software with a social content planner, and its AI is metered: its page caps the Free tier at up to 200 uses of standard AI tools. It has no real-estate features, and what keeps weekly marketing alive is a repeatable calendar slot, the same argument as the tasks worth automating first.

What these AI tools for real estate agents actually cost

Prices are from vendor pricing pages, 6 August 2026. Every tool name links to the page it was read from.

Four items above publish no price: Lofty, Ylopo, Canva Enterprise and the Canva AI Pass add-on. Two of them are the platforms built for your trade. Ylopo’s stated reason: “Posting one number would overcharge some and shortchange others.” Its hero claims “75K+ Agents on the platform” while a pop-up says “Trusted by 2,200+ agents & teams.” A percentage of ad spend, most of what Lofty publishes, is a different commitment from a subscription: the bill grows with the budget.

Start with the copy job: ChatGPT Plus at $20 a month on listing descriptions and follow-up emails, and stop writing those two by hand. It is the only listing-copy option here with a published price, which is why you begin there. Add lead response second, once you can say how many leads you dropped last month. Leave the platforms until you have a number to hold them to. The general argument is choosing tools by the job.

See what your marketing is actually doing

Compliance: what you are still on the hook for

None of this is legal advice, and rules vary by state and by MLS. Read your own MLS’s rules and ask your broker.

The advertisement is yours, whoever wrote it. 24 CFR 100.75(a) makes it unlawful to publish an advertisement for a dwelling “which indicates any preference, limitation or discrimination” on the protected bases, and paragraph (b) covers “written or oral” statements, which puts an AI voice agent in scope. The rule names those bases: race, color, religion, sex, handicap, familial status and national origin. Familial status catches listing copy most often. A footnote to HUD’s 2024 guidance below kills the “the AI wrote it” defense: courts read the Act to mean a defendant can violate it if the ad “indicates discrimination to an ‘ordinary reader’ or ‘ordinary listener,’ regardless of whether the defendant intended to discriminate.” An ordinary reader means the average person who sees your ad, and your own reading does not decide it.

Targeting can discriminate without you knowing. HUD’s 2024 guidance uses a real estate agent as its example: upload a “custom audience” (a list you give the ad platform so it targets those people and others like them) built from open-house attendees, and if all of them were White, “the new home ad will be eligible to be delivered only to White consumers.” Treat it as 2024 guidance rather than current enforcement posture: it sits on archives.hud.gov, marked “Content Archived: February 3, 2025”, and quotes an Acting Secretary, though the regulation it interprets is current. Its four recommendations: use ad platforms that tell you how they manage discriminatory delivery, flag housing ads as housing, check where an uploaded list came from, and, in HUD’s words, “Monitor outcomes of advertising campaigns for housing-related ads, to the extent possible, to identify and mitigate discriminatory outcomes.”

Altered photos have named rules, and where you work decides which. Fair housing and the robocall rule below are federal and reach every agent. The image rules are local. California AB 723, signed on 10 October 2025, adds section 10140.8 to the Business and Professions Code: a California licensee using a digitally altered image in an advertisement must include “a statement disclosing that the image has been altered” plus a link, URL or QR code identifying the original image. The definition names artificial intelligence, and exempts ordinary lighting, cropping and color correction.

The statute names no effective date and no dollar penalty. Two MLSs have set their own dates. San Diego MLS enforces it as an MLS rule from 1 January 2026, and notes that AB 723 itself reaches all your advertising, websites and social media included. SDMLS keeps you liable “even when using third-party vendors, photographers, or service providers”. ARMLS, Arizona’s regional multiple listing service, requires a Disclosure of Digitally Altered Images from 28 May 2026. Both rules govern altered images, and no MLS rule turned up requiring you to label AI-written text.

An AI voice on the phone is a robocall for consent purposes. The FCC’s February 2024 Declaratory Ruling recognized that AI-generated voices are “artificial” under the Telephone Consumer Protection Act, pulling them inside the existing rule requiring “prior express written consent from consumers before robocalling them”. Consumers can sue in court.

The Code of Ethics does not change with the tooling. Article 12 of the 2026 Code requires that recipients of real estate communications “are, or have been, notified that those communications are from a real estate professional”, which a bot answering your leads has to clear.

See how we run the whole engine

What stays yours: pricing, negotiation and the CMA

A comparative market analysis, the priced comparison of recent sales you build to tell a seller what their home is worth, is the clearest case. A model can assemble one, and you defend it at the listing appointment to a seller with their own number. The counter-offer conversation runs on what you heard in the other agent’s voice, and no model was on that call. The fair housing read on every AI-drafted description is the step you cannot hand off. NAR’s Legal Affairs page puts the wider risk in its own words, typo included: “AI platforms are not 100% accurate and its output may not comply with fair housing laws.” What not to hand to AI goes deeper.

Where a done-for-you engine fits instead of ten subscriptions

The honest version of a full stack, on one monthly line: a CRM with calling at about $108 per user, ChatGPT at $20, Canva Pro at US$15 on its annual price, and 200 minutes of AI calling at $14 to $62. That is roughly $157 to $205 a month for one agent, before virtual staging at US$24.00 an image. Then add the hours to check every output against fair housing language and your MLS’s image rules.

That total is where a managed service starts to look cheaper. We run the whole engine for clients: content, follow-up and weekly publishing, with a person reviewing every output before it goes live. If you want the AI tools for real estate agents question off your desk, our managed marketing pipeline is where the stack becomes our problem. Start with the free marketing assessment.

Frequently asked questions

Can AI write real estate listing descriptions?

Yes, and 46 percent of respondents in NAR’s 2025 survey reported using AI-generated content for listing descriptions. You remain the advertiser under 24 CFR 100.75, so every draft needs a fair housing read against the seven protected bases. No MLS rule found here requires labeling AI-written text, though two require disclosure of altered photos.

Is AI worth it for a solo agent?

One test decides it. If you are dropping work you have already been paid to do, buy ChatGPT Plus at $20 a month and point it at listing copy and follow-up emails. If nothing is being dropped, buy nothing yet. NAR’s own member survey, on a 2.5 percent response rate, found 46 percent saying AI had no noticeable impact.

What do AI tools for real estate agents cost per month?

From $0 to $20 for a general assistant like ChatGPT. Follow Up Boss lists Grow at $69 per user per month plus $39 for calling. Voice agents run $0.07 to $0.31 a minute, roughly $14 to $62 for 200 minutes. Virtual staging is US$24.00 an image. Lofty and Ylopo publish nothing.

Will AI replace real estate agents?

Nothing in the available evidence points that way. BLS projects 3 percent employment growth for brokers and sales agents from 2024 to 2034, and every performance claim in this category traces back to a vendor’s own marketing.