Skip to main content

What Real Estate Agents Should Never Automate With AI

Written by Coach Emily on . Posted in Uncategorized.

By Emily Terrell — Top Coach and Speaker at Tom Ferry International. Speaker for NAHREP, eXp Con, and brokerages nationwide.

Real estate agents should never automate five things with AI: opt-out judgment, live answers about neighborhoods or schools, advertising approval, review filtering, and messages triggered by a client’s life events. Each one reaches a consumer with no human between the trigger and the send. This guide gives you the test, the five zones, and the safer setup.

Key Takeaways

  • Drafting with AI and automating with AI are different risks: a draft waits for you, an automation doesn’t.
  • The test is one question: is there a human between the trigger and the send?
  • Five workflows fail that test every time: opt-out judgment, live neighborhood answers, ad approval, review filtering, and life-event triggers.
  • Automate the schedule, the routing, and the reminder. Keep the decision.
  • Your CRM vendor’s compliance features help, but the consent record and the advertisement are still yours.

What does it mean to automate with AI in real estate?

Automating with AI means a system acts on a trigger, without a person approving that specific action. The action might be sending a text, posting to social media, replying to a lead, or tagging a contact. Drafting is different. AI produces the words, and you decide whether they go anywhere.

The question of what agents should never use AI for at all is about the output. This post is about the trigger.

That distinction matters because most of the compliance problems I see in coaching aren’t bad drafts. They’re decent drafts that went out at 11:40 p.m. to the wrong person, because a rule someone set up eight months ago said so.

Why this matters for real estate agents

AI is already in the workflow. According to NAR’s 2025 Technology Survey (September 2025), 20% of agents use AI tools daily and another 22% use them weekly, yet 46% said AI had no noticeable impact on their business (NAR, 2025).

Agents know the risk is there. According to RPR’s February 2026 survey of 225 NAR members, 82% of agents currently use AI. In the same survey, 49% named compliance or legal issues as a top concern, and 28% named Fair Housing (RPR, 2026).

Put those together. Most agents use AI, half of them worry about compliance, and almost none have mapped which of their automations fire without a human. A bad draft makes one mistake. A bad trigger makes the same mistake for every contact it touches.

“Every automation is a decision you made once and then stopped watching. Before anything fires to a consumer, I want one person who can tell me what it says, who it goes to, and how it stops.”
— Emily Terrell, Tom Ferry Coach

Is there a human between the trigger and the send?

Run every automation in your business through that one question. If the answer is yes, AI can draft, sort, score, and schedule all day. If the answer is no, the workflow needs a hard stop built in, or it stays manual.

For the task-by-task version of this split on a team, see what to delegate to AI vs. a human. The five categories below fail the test no matter how good the tool is.

The five things real estate agents should never automate

Should real estate agents automate opt-out handling?

Automate the suppression, but never the judgment. A keyword filter that catches “STOP” is fine, and it’s fast. The problem is everything a keyword filter misses.

Real people don’t text STOP. They text “we already bought, take me off this” or “my mom passed, please stop sending these.”

Under current FCC rules, a consumer who revokes consent to your marketing texts or calls revokes it for all of your marketing robocalls and robotexts. You have no more than 10 business days to honor it.

The FCC changed part of this in October 2026. The new order relaxes the rules for informational messages and lets senders designate specific opt-out methods if they’re clearly disclosed. Marketing messages stay under the revoke-all standard, and the Commission is taking comment on shortening the deadline to seven business days (Squire Patton Boggs, 2026).

What works: any inbound reply that isn’t a clear yes pauses the sequence and creates a task for a person the same business day. AI can flag likely opt-outs. A human confirms them. The full consent framework is in AI texting rules for real estate agents.

Can an AI chatbot answer buyer questions about neighborhoods and schools?

Not live and unattended. Picture a chatbot answering three questions at 2 a.m.: “Is this a safe area?”, “How are the schools?”, and “Is this neighborhood good for families?” Every answer creates Fair Housing exposure in real time, with no chance for review.

The Fair Housing Act prohibits housing statements that indicate a preference or limitation based on a protected class. It doesn’t care whether a person or a model typed them. A trained agent redirects those questions to objective sources. A model trained on decades of listing copy just answers them.

What works: automate the handoff, not the answer. The bot can schedule a showing, confirm a listing’s MLS facts, and capture the question. Anything about safety, schools, demographics, or “who lives here” routes to a licensed human, along with a standard redirect to public data sources.

Can AI auto-publish listing posts and ads?

Automate the calendar, not the approval. Scheduling a post a human already approved is fine. When AI generates and publishes on its own, broker identification disappears.

In Texas, TREC Rule 535.155 requires every advertisement to include two names: the license holder’s or team’s name, and the broker’s name. The broker’s name must be at least half the size of the largest agent or team contact information. TREC’s definition of advertising includes email, text messages, and social media (TREC). AI tools leave the broker’s name out every time because they don’t know the rule exists. Agents outside Texas need their own commission’s equivalent rule.

What works: AI drafts a batch of posts. A person approves the batch against a three-point check: property facts match the MLS record, the broker’s name is present, and there’s no protected-class language. Only approved posts enter the scheduler.

Should agents automate review requests and testimonial displays?

Automate the ask, but never the filter. Sending every closed client the same review request on day three is a good system. Filtering is different. That means routing only happy clients to the public review page, or auto-publishing only five-star reviews on a page that implies it shows everything.

The FTC’s final rule on consumer reviews and testimonials was announced August 14, 2024. It bars a business from claiming the reviews on its website represent all or most of the reviews submitted when it has suppressed reviews based on rating or negative sentiment. The same rule covers AI-generated fake reviews and allows civil penalties against knowing violators (FTC, 2024).

What works: send one request, with the same wording, to every closed client. A human decides how reviews are displayed. If you curate, the page says so.

Should automations fire off life events or deal status?

No. No statute covers this one, and it does the most damage to relationships. CRM triggers built on transaction stages assume the data is current, and it usually isn’t. Some examples:

  • A “Happy 1-year home anniversary” email goes to a buyer whose contract terminated, because nobody updated the stage.
  • A price-improvement blast goes to your sphere before the seller approved the messaging.
  • A “Congratulations on closing!” text goes out on the day the closing slipped.
  • An automated market update keeps running on a past client who just told you about a divorce.

What works: any automation tied to a transaction stage, a price change, or a personal event needs a human to confirm the trigger is true before it sends. That’s a thirty-second task.

This is general information, not legal advice. TCPA, FTC, Fair Housing, and state license rules vary and change. Consult your broker and an attorney licensed in your state before setting policy.

What should real estate agents automate instead?

Plenty. The goal isn’t fewer automations. It’s automations that stop at the right place.

  • Speed-to-lead: an instant acknowledgment to contacts with documented consent, then a human takes the second touch
  • Task creation: every new lead, every reply, and every stage change creates a to-do for a named person
  • Transaction timeline reminders for your own team
  • Draft generation: listing copy, captions, newsletters, and follow-ups queued for approval
  • Post scheduling for content a human already approved
  • Review requests sent the same way to every closed client
  • Data cleanup: duplicate merges, missing-field flags, and stale-lead alerts

Notice what these share. Each automation moves work to a person instead of replacing that person’s decision.

Common mistakes

Assuming the vendor’s compliance page covers you. Vendor features help. The consent record and the advertisement still belong to you.

Building automations on stale CRM data. A trigger is only as accurate as the stage, tag, or date it reads. If nobody owns data cleanup, every stage-based automation is a guess.

Treating “STOP” as the only opt-out. Real opt-outs arrive as full sentences. If your sequence doesn’t pause on any reply, it will text someone who already asked you to stop.

Letting the chatbot answer neighborhood questions on its own. The bot’s job is to capture and route the question. Answers about safety, schools, or who lives in an area belong to a licensed person.

Auditing drafts but never auditing triggers. Once a quarter, pull a list of every active automation in your CRM, email platform, and social scheduler. Name an owner for each one. Anything without an owner gets turned off.

Frequently Asked Questions

What should real estate agents never automate?

Five workflows should never run without a human between the trigger and the send: deciding whether a reply is an opt-out, live answers about neighborhoods, schools, or safety, publishing advertising without approval, filtering which reviews get shown, and messages triggered by deal status or personal life events. Everything else can be automated if a person owns the decision.

What’s the difference between AI drafting and AI automation?

Drafting means AI produces content and a person decides whether it goes anywhere. Automation means a system acts on a trigger — sends, posts, replies — without a person approving that specific action. Drafting risks are caught in review. Automation risks repeat across every contact the trigger touches, which is why automated workflows need hard stops that drafts don’t.

Is it legal for real estate agents to send automated texts?

Yes, with the right consent. Automated marketing texts require prior express written consent under the TCPA, and you must honor opt-outs within no more than 10 business days under current FCC rules. The consent record sits with you, not your CRM vendor. This is general information, not legal advice; confirm your setup with your broker and an attorney.

Can an AI chatbot on my website answer buyer questions?

It can answer factual listing questions, schedule showings, and capture contact details. It shouldn’t answer questions about neighborhood safety, schools, or who lives in an area, because those answers can indicate a preference based on protected class under the Fair Housing Act. Route those to a licensed agent with a standard redirect to objective public data sources.

Should I automate review requests to past clients?

Yes — send the same request, with the same wording, to every closed client. What you shouldn’t automate is the filter. The FTC’s 2024 rule on reviews and testimonials bars misrepresenting that the reviews on your website reflect all or most of those submitted when you’ve suppressed reviews based on rating or negative sentiment.

What real estate tasks are safe to automate with AI?

Speed-to-lead acknowledgments to consented contacts, task creation for your team, transaction reminders, draft generation queued for approval, scheduling of pre-approved posts, uniform review requests, and CRM data cleanup. The common thread: the automation moves work to a named person instead of replacing that person’s decision. If a human can stop it before a consumer sees it, it’s a candidate.

Do brokerages need an automation policy?

Yes. When agents run automations through their own CRMs, the brokerage ends up with dozens of unreviewed triggers sending under its name. A one-page policy should list what never runs unattended, require a named owner for every active automation, and set a quarterly audit. Brokerage leaders should confirm specifics with counsel.

Bring this to your team or event

Emily Terrell speaks at brokerage events, real estate conferences, and team trainings on AI, systems, and social media — the exact playbook in this post, delivered live to your audience. As a Top Coach and Speaker at Tom Ferry International and an active agent closing 70+ transactions a year, Emily speaks from the stage about what’s working right now, not theory. Recent stages include NAHREP and eXp Con.

Book Emily to speak at your next event:
Email: eterrell@yourcoach.com
Phone: (210) 400-9191
Web: coachemilyterrell.com

For real estate agents who want to implement this: Get the weekly real estate prompt library at weeklyrealestateprompts.com or follow @coachemilyterrell on Instagram for daily systems and AI breakdowns.