I was at a conference talking with a gentleman who ran a sizable mortgage organization. The conversation turned to technology, his loan officers, and the CRM he wanted to build.
As he described what his team needed, I started walking him through how I use AI agents in my own business.
Checking email across accounts. Reviewing my calendar. Researching people before birthday calls. Gathering information about new for-sale-by-owner listings. Preparing follow-up and recording conversations in my CRM.
He was surprised by how much of this work I was already assigning to agents.
The conversation reminded me how different using AI looks from one business to another. Someone might use ChatGPT to write an email while another person uses agents to prepare a daily call list and update contact records after the calls.
For a Realtor thinking about hiring help, those differences deserve attention.
Before you write the next job description, look at the work you need completed. How much involves gathering information, checking records, or preparing something for your review?
Those are the assignments I would evaluate first.
Before calling a past client, you look up the last conversation. Before contacting a homeowner, you research the property. Before starting your day, you check email and review appointments.
Each task takes time. Together, they compete with showings, negotiations, and conversations.
I know how to do all of those jobs. My agents help me get the preparation done so I have more time to talk with people.
The same thinking guided me when I built my virtual assistant company. Find repeatable work. Explain the process. Assign responsibility. Review the result.
With AI agents, I follow a similar approach.
Years ago, I built a process where virtual assistants checked birthdays and helped me send personal messages. I’ve used versions of this approach for about a decade.
When I started with Grok Bot, I wanted to take the research further.
My question was simple: How do I make this person feel special on their birthday?
A text might be appropriate. For someone close to me, I might prefer a phone call. Another person might appreciate a short video recorded specifically for them.
To make a good decision, I wanted context.
When did we last speak? What did we discuss? Had they bought a house recently? Had they shared something about their family?
A birthday gives me a yearly reason to reconnect and catch up.
Some days, I have around 30 birthdays to review. Researching every person myself would take a substantial part of the day.
So I assigned the preparation to an agent.
The result I want is a useful summary of our relationship, drawn from available information and past interactions. Then I decide how to reach out.
I don’t want a meaningful relationship reduced to an automatic greeting. I want the research ready so I’m prepared to have a personal conversation.
A name and an avatar don’t train an assistant.
You need to explain the assignment, the information available, and the result you expect.
For my birthday system, finding names is only the beginning. I want context to help me choose an appropriate message or call.
For a Realtor’s follow-up system, “help with leads” is too broad. A more useful assignment would be:
“Prepare a daily follow-up list from my CRM. Include the last recorded conversation, any promised next step, and the reason each person needs attention. Flag missing information for review.”
Now you have a result to evaluate.
Did the bot choose relevant contacts? Did the summary reflect the conversation? Did the work save preparation time?
Work through those questions before adding another assignment.
I’ve built around 20 agents, but my approach is to develop one specialty at a time. Each needs a defined purpose and a process I understand.
Past-client follow-up is one of my regular uses.
My personal target is to reach about 10 past clients a day through calls, texts, or another appropriate form of contact.
The preparation matters. I want to know where we left off and whether there’s something worth revisiting.
Afterward, I record a recap for the agent.
Someone answered a text. Someone didn’t respond. I had a longer conversation with another person who mentioned buying a rental property in the future.
In my setup, the agent takes those details and adds them to the CRM.
Now there’s information to support the next conversation. When I review the relationship later, I have something more useful than a name and phone number.

More Closings. Fewer Financing Surprises.
Tired of deals dying because your buyer's lender dropped the ball? Keith is the mortgage partner listing agents call when they need deals protected, difficult files approved, and buyers who show up pre-qualified.
This also shapes how I think about the future of CRMs. I expect more of my interaction with those systems to happen through agents: asking for information, recording updates, and reviewing results.
The records still matter. I want to spend less time entering and searching them.
During one past-client conversation, I learned the client’s husband, a veteran, was dealing with health problems and struggling to pursue help through the VA.
I knew someone who helped veterans work through their benefits. I connected them.
Later, I learned he was receiving approximately $4,900 a month in disability benefits.
My contribution was making the connection. The benefits specialist handled the benefits work.
There was no mortgage application involved. A conversation revealed a need, and I knew someone who helped.
This is why I care about making time for calls. When your only contact is an automated message, you lose opportunities to hear what’s happening in someone’s life.
Another agent assignment in my business involves for-sale-by-owner listings.
We use listing sources which update as new properties appear. The agent gathers available property and ownership information and organizes the findings for review.
I want the preparation completed before the call.
For a Realtor, this gives you a starting point for asking informed questions. You still need to confirm details and listen to the homeowner’s explanation.
Research also doesn’t excuse a misleading introduction.
At an open house, I spoke with a woman selling her own home. She described repeated calls from agents claiming to have a buyer. Those conversations hadn’t produced the buyers she expected, and she felt the approach was deceptive.
Her reaction was straightforward: she didn’t want to work with those agents.
If you have buyers whose criteria potentially fit a property, explain exactly what you know. Don’t turn a possible fit into a claim of committed interest.
An agent should help you prepare for an honest conversation.
Give a dedicated writing bot several examples of your listing descriptions. Explain what sounds like you and what you want changed.
Then provide verified details about the next property.
The bot prepares a draft. You review the facts, adjust the language, and approve the final version.
This is a useful early assignment because the result is easy to inspect. You know the property, you know your voice, and you know how much editing the draft requires.
The same principle applies to email and calendar reviews. Define what deserves attention and what you expect the agent to bring back.
Keep the responsibility specific enough to evaluate.
Once several agents have reliable assignments, a coordinating agent becomes more useful.
A “chief of staff” might summarize completed work, identify missing information, and bring you items needing a decision.
But build the individual processes first.
You’ll learn more from improving one daily follow-up list than from creating eight bots with vague responsibilities.
Review the work. Correct misunderstandings. Refine the instructions. Then decide whether the next assignment belongs with AI, your virtual assistant, your transaction coordinator, or you.
This is the same question I explore in “Your TC, VA, and AI Are Probably Doing the Wrong Jobs.” Every task needs an appropriate owner.
Before hiring another assistant, write down the recurring work creating the pressure.
Choose one task with a clear result. Test how much preparation an agent handles and how much review you still need to provide.
Track the cost, too. In “My AI Agent Burned Through $329 in Four Days,” I shared why agent usage deserves attention alongside the subscription price.
Your decision should reflect completed work, correction time, and expense.
For me, the benefit is more room for conversations. Calls with past clients. Better-prepared outreach. Personal birthday messages. Time to hear what someone needs.
Let AI prepare the research. You make the calls.
My son, Jake, and I are developing a business to help professionals set up AI agents. I’m bringing my experience with delegation and business systems, and he’s learning the implementation work while building a business to help fund college.
We’re offering a free initial review of the work you want help handling.
Send me one recurring task, how you complete the work today, and what you want ready when you sit down.
We’ll start with a specific job and define what a useful result looks like.
Start with one repeatable preparation task and define the result you want back. The article's example is a daily follow-up list with the last conversation, any promised next step, and missing information flagged for review.
No. Use the agent to gather available information and prepare context. The Realtor decides how to reach out, what to say, and how to respond to the person.
An agent can prepare the call by summarizing the relationship, identifying the last recorded conversation, and recording a recap in the CRM after the Realtor provides the details.
An agent can gather available property and ownership information before the call. The Realtor still needs to confirm the details, listen to the homeowner, and make an honest introduction.
Add a coordinating agent after individual assignments have reliable processes. The article recommends building specialists first, then using a manager to summarize completed work and bring back decisions that need attention.
Keith has helped 2,980+ families close on homes across 12+ markets over 24+ years, with $310M+ in closed loans. He specializes in difficult files, listing-agent partnerships, and the kind of complex deals most lenders turn away.
He writes the Realtor Money Alerts to give agents the plays, programs, and systems that protect deals and generate more closings.
Drop your email and I'll send you the next one first.
If you want to see if we can get more buyers and sellers for you, a 15-minute call is all it takes. No pitch. No pressure. Just a real conversation about your business.
Keith works with a small number of agents in each market. If there is a fit, you will know by the end of the call.
"Keith helped us get an offer on a $2.8 million dollar listing in Arrington, TN with a custom funnel. Incredible results."
"Keith has been instrumental in the growth of our listing leads coming in. He is a true partner."
We help you grow your business and get you more clients · Keith Goeringer NMLS #488023 · No obligation