Real estate runs on writing and follow up: listings, social posts, buyer emails, and the 40th text to a lead who went quiet in March. AI handles most of that writing well. It can also write a listing that breaks fair housing rules, with your license number at the bottom. This article covers where AI helps an agent or a small brokerage, and the checks that keep it out of trouble.

Start with listing descriptions

A good listing description takes most agents 30 to 60 minutes. With AI it takes about 10, including the edit. Give the chatbot the facts only: beds, baths, square footage, lot size, year built, upgrades with dates, HOA details, and what the seller said they love about the house. Ask for 150 to 250 words in a plain tone, plus a 50 word version for portals and social posts.

Use a business account. ChatGPT, Claude, and Microsoft Copilot all run roughly $20 to $30 per user per month on business plans, which keep seller details out of model training by default. Many MLS and CRM systems include a description writer too, so check what you already pay for.

Action: write your next listing with AI from a fact sheet, and time it against your usual method.

Check every draft for fair housing language

The Fair Housing Act bars advertising that states or implies a preference based on race, color, religion, sex, disability, familial status, or national origin, and many states and cities add more. The agent and broker are responsible for what gets published, whoever wrote it.

AI drafts drift toward phrases common in old listings. Watch for:

  • Who the home suits. "Perfect for young families," "ideal for empty nesters," "great bachelor pad."
  • Who lives nearby. "Safe neighborhood," "exclusive community," "quiet, mature residents."
  • Places of worship and schools as selling points about people. "Walk to St. Mary's" reads differently from "0.4 miles to the train station."
  • Ability. "Ideal for active buyers" or "no wheelchairs." Describe the property, such as "single level, no steps at entry," not the buyer.

Put a standing instruction in your prompt: "Describe the property and its features only. Do not describe the ideal buyer, the neighbors, or the neighborhood's residents." It reduces the problem. It does not remove it, so a person still reads every draft.

Action: add a fair housing word check to your listing checklist and require it before anything goes live.

Disclose virtual staging and edited photos

AI photo tools can stage an empty room for a few dollars a photo, or remove a power line from a front shot in seconds. Many MLSs require you to label virtually staged images, and some states now require disclosure of digitally altered photos in real estate advertising. Removing a permanent feature such as a cracked driveway can become a misrepresentation claim.

Action: label every staged or edited photo, keep the original in the listing file, and never edit out something a buyer would see on a showing.

Use AI for lead follow up, with limits

CRMs built for real estate, such as Follow Up Boss, Lofty, or kvCORE, typically cost roughly $60 to $500 a month depending on team size, and most now offer AI that drafts replies or runs text conversations with new leads.

Two limits matter. First, automated texts and calls fall under the Telephone Consumer Protection Act, and the FCC has said AI generated voices count as artificial voices under that law. Get clear consent on your lead forms before an automated system contacts anyone. Second, AI should not answer questions about schools, crime, or "what kind of people live here." Those go to a person, who points to public data sources.

Action: review the consent wording on your website forms and portal lead sources this week, and set your CRM so AI drafts are approved by an agent for the first 30 days.

Save time on the paperwork around the deal

AI works well on the text that surrounds a transaction. Summarizing a 40 page inspection report into a one page list for the buyer. Turning showing feedback into a weekly seller update. Drafting a timeline email after contract with every deadline from the purchase agreement. An agent who handles 15 to 25 deals a year can save two to four hours a week here.

Keep the deadlines and dollar figures tied to the signed documents. Check every date the AI writes against the contract, because a wrong option period or financing deadline costs far more than the time saved.

Action: pick one recurring email, such as the weekly seller update, and build a saved prompt for it.

Brokers: write the rules once

In a brokerage with 5 to 50 agents, each agent will otherwise pick their own tools and habits. A one page policy covers it: which tools are approved, that client and seller data stays in business accounts, that every listing gets a fair housing check, that edited photos are labeled, and that automated contact needs documented consent.

Action: draft the one page policy this month and have each agent acknowledge it.

If you run a team or a small brokerage and want help choosing tools and setting up the checks, the first call is free. Bring a recent listing and your lead follow up process, and we will show you what we would hand to AI and what we would keep with your agents. Get in touch and we will find a time.