← Back to Blog

AI Automation for Real Estate Agents: How One Team Went From Seven Agents to Three and Kept Growing

A Columbus team lost four agents and replaced the busywork with AI instead of new hires. Here are the four systems worth copying, the order to build them in, and what they cost a month.

Industry10 min read•

Seven agents to three, and the GCI went up

In 2019 the LKW Home Team in Columbus, Ohio had seven agents. By 2023 it had three. Most teams that lose four people shrink. This one reported a 35.8 percent rise in gross commission income with three agents, according to Real Estate News' March 2026 coverage of co-owner Lauren Lucas, a Keller Williams-affiliated agent who built AI systems to do the work the departed agents used to do.

You do not need her exact setup. You need to know which jobs she handed to software, in what order, and where a person still has to take over. That is what this guide covers, with real monthly costs and the two compliance traps that catch agents who automate texting and messaging.

What the Columbus team actually automated

Real Estate News described four pieces. A ChatGPT-based agent that writes and posts social content. A voice AI that had answered more than 900 inbound calls and hands the caller to a human when it is less than 95 percent confident. QR code sign-in forms at open houses that capture contact details and consent. And CRM automation that follows up from there. The article says the setup frees 10 to 20 hours a week for Lucas and her partner combined. A podcast episode she appeared on lists ManyChat, GoHighLevel and ChatGPT among her tools.

The case study on this site, How One Real Estate Agent Grew to $400K GCI Using AI Automation Instead of Hiring, reports larger figures, including a GCI increase of over $400,000 and 75 percent of operations automated. We could not trace those numbers to a primary source, so treat the press figure of 35.8 percent as the solid one. Either way, the pattern is the useful part: the team did not buy an all-in-one AI platform. It automated the four places where leads were getting lost.

The four systems worth copying

Every small team leaks leads in the same places. The open house sign-in sheet that nobody types up until Tuesday. The Instagram comment asking for the price that gets a reply two days later. The Zillow or website lead who gets a call four hours after they filled in the form, by which time they have talked to someone else. And the social posting that stops the moment the team gets busy, which is exactly when it matters.

Each of these is a repetitive, rules-based job, which is what automation does well. None of them requires the AI to negotiate, give an opinion on a neighborhood or advise on price, which is what it should never do. The voice receptionist is the fifth piece; we cover it separately in AI Receptionist for Real Estate, so this guide focuses on the other four.

Four systems: what the automation does, what you keep System Automation handles You keep Open house QR Sign-in, consent, CRM entry, first text The follow-up call Instagram DMs Keyword reply, 3 qualifying questions Any real conversation Speed to lead Text in a minute, book on calendar Price, offers, neighborhoods Content drafts Weekly posts from your notes Edit and press post Build them top to bottom. Each one feeds the CRM the next one depends on.

A worked example: a two-agent team in a mid-size market

Picture a two-agent team doing about 30 sales a year, mostly from referrals. They hold two open houses a month with about 15 visitors each, get around 40 comments a month on listing reels, and receive 20 web leads a month from their site and portals.

Before automation, maybe half the open house visitors get a follow-up, the Instagram comments get a reply when someone remembers, and web leads wait hours. After the four systems are running, every open house visitor gets a thank-you text within minutes of signing in, Instagram commenters who type TOUR get a DM that asks three questions and offers a showing time, and web leads get a text within a minute asking when they would like to see the home. The agents start their day with a list of people who replied, not a pile of names to chase.

What does that add up to? We cannot promise a number, and nobody honestly can. If faster follow-up turned even two of those leads a year into closings, at a typical commission that pays for the stack for several years. This is an illustration, not a documented team.

Step by step: build it in this order

Step 1: Find where your leads die

Export your last 50 leads from your CRM or inbox and note, for each, the source, how long until first contact and what happened. You will usually find one source with the slowest response and the most lost leads. Build that system first. If you skip this step you will automate what is easy instead of what is costing you deals.

Step 2: Set up one CRM pipeline and register for texting

Everything below needs a place to land. GoHighLevel is what the Columbus team used; its Starter plan is $97 a month. Create a pipeline with stages such as new, contacted, appointment set, showing done, under contract. Before any automated text goes out, complete A2P 10DLC registration. HighLevel's help center says this is required to text US numbers from a standard local number and to wait for verified status before sending. Unregistered texts get filtered or blocked.

Step 3: Replace the paper open house sheet

Make a short form in your CRM with name, phone, email, whether they have an agent, and a consent checkbox with plain wording about receiving texts, how often, and that they can reply STOP. Print a QR code to it on a sign by the door. Set a workflow: when the form is submitted, add the contact to the pipeline, tag it with the property address and send a text within two minutes thanking them and asking one question, such as what did you think of the kitchen. Replies go to your phone.

Step 4: Turn Instagram comments into qualified conversations

In ManyChat, connect your Instagram professional account and create an automation with a comment trigger on your listing posts for a keyword such as TOUR. ManyChat's help center describes comment triggers that can watch one post or all posts and match keywords. The DM flow should send the listing link and ask three questions with buttons: buying timeline, whether they are pre-approved, and whether they already have an agent. If they ask for a showing, collect a phone number with consent and send the contact to your CRM through ManyChat's Make or Zapier integration. Meta's messaging policy gives businesses a 24-hour window to reply after a person's message, and requires a way to hand off to a human, so set the flow to alert you when someone types a free-text question.

Step 5: Text every web lead within a minute

In your CRM, create a workflow that fires when a web or portal lead arrives. Send a short text from your name, not a company name: hi, it is Sam, I saw you asked about the house on Elm, would a showing tomorrow at 5 or Saturday at 11 work? Include a booking link to your calendar. If the lead replies with anything other than a time, notify you immediately. If you use GoHighLevel's AI Employee to hold the conversation, restrict it to scheduling and basic listing facts.

Step 6: Write the rules the AI must never break

In every AI prompt or bot setting, add explicit limits: do not discuss price strategy, offers or negotiation; do not describe neighborhoods, schools, safety or who lives there; do not say anything about a property that is not in the listing. For neighborhood questions, reply that you will send the public data and alert a person. The Fair Housing Act applies to what your bot says just as it applies to you, and steering complaints do not care that software typed the message.

Step 7: Draft content from your own notes

Create a ChatGPT project or Claude project with your brand notes, your market, your past best posts and the things you never say. Every Monday, paste in that week's raw material: a new listing, a closing, a market stat from your MLS, a client question. Ask for five posts in your voice. Edit them and post them yourself or schedule them. The Columbus team let an agent post on its own; for most agents, a human press of the publish button is cheap insurance.

Step 8: Review the numbers every Friday

Check four things: median time to first contact, number of replies, appointments set and showings done, broken down by source. If one source has replies but no appointments, the questions or the booking offer need work. Adjust one thing a week.

What this costs

As of November 2026: GoHighLevel Starter is $97 a month and Unlimited is $297 a month, per its pricing page. Its AI Employee add-on is $50 a month per sub-account on the lower plan and $97 on Unlimited, and conversation AI usage is billed on top. ManyChat Pro is $39 a month billed monthly or $29 a month billed annually for up to 2,500 active contacts, per ManyChat's help center. ChatGPT Plus and Claude Pro are each $20 a month. Texting registration and message fees are small but not zero.

A typical stack of GoHighLevel Starter, ManyChat Pro and one AI subscription runs about $150 to $160 a month before message fees. Budget a weekend to build the first two systems and an hour a week to review them. Compare that to a part-time assistant, and remember the assistant can also do things software cannot, like run to the printer before an open house.

When not to do this

Skip the Instagram flow if you post rarely; there is nothing for it to catch. Skip GoHighLevel if your brokerage already gives you a CRM that can send texts on a trigger, and build the same workflows there. Do not automate anything for a team where nobody will answer the replies, because fast first texts followed by slow human follow-up is worse than no automation. And if your brokerage has rules about AI tools, advertising or texting, those come first.

Frequently Asked Questions

What can AI automate for a real estate agent?▾

The repetitive parts of lead handling: open house sign-ins, first replies to web and social leads, scheduling showings, follow-up reminders and draft social posts. Pricing advice, negotiation and anything about neighborhoods should stay with a licensed person.

How much does AI automation cost a real estate agent?▾

A common small-team stack of GoHighLevel Starter at $97 a month, ManyChat Pro at $29 to $39 a month and a $20 AI subscription comes to about $150 to $160 a month as of November 2026, plus texting fees.

Is it legal to text leads automatically?▾

Yes, with consent and registration. Collect clear consent on your forms, include opt-out instructions, and register for A2P 10DLC before sending from a local number, which HighLevel's help center says is required for business texting to US numbers.

Can ManyChat automate Instagram DMs for realtors?▾

Yes. ManyChat can reply by DM when someone comments a keyword on a post, ask qualifying questions with buttons and pass contacts to a CRM. Set it to hand off to you whenever someone asks a question the flow cannot answer.

Will AI replace real estate agents?▾

Not the parts clients pay for. The Columbus team in this guide used AI to replace busywork and kept agents for showings, pricing and negotiation. What it did change is how many people a small team needs to keep up with leads.

Does fair housing law apply to AI chatbots?▾

Yes. Anything your bot says is your advertising and communication. Tell it never to describe neighborhoods, schools, safety or the people who live in an area, and to hand those questions to a person who can point to public data.

Keep Reading

Explore Related Niches

Related Strategies

Real EstateReal Estate AgentsFreelance/Agency

A real estate agent grew to $400K GCI by building AI systems that run the business without a full team.

BeginnerUnder $50/moVery High ($10K+/mo)
View Strategy
Lead GenReal Estate AgentsFreelance/Agency

A first AI automation agency client worth $20,000 came from a casual networking conversation, not cold outreach or paid ads.

IntermediateUnder $50/moVery High ($10K+/mo)
View Strategy
Real EstateReal Estate AgentsInternal Tool

An Obsidian knowledge base connected to Claude Code that gives the AI full context on a three company, 50 plus property real estate operation from the first message.

BeginnerFreeHigh ($2K-10K/mo)
View Strategy

Keep Exploring the Directory

Browse more case studies, compare tools, or dig into the full tool stack behind these workflows.

Want more strategies like this?

Get weekly AI agent case studies in your inbox.