Most real estate teams don't worry that AI will be too slow. They worry it will make them sound like everyone else. If your clients chose you because someone calls them back, explains every document, and never lets them feel forgotten, handing that work to software can feel like giving away the reason they hired you.
Carrie Hensley lives with that tension every day. She is Director of Business Operations at SCH Real Estate in Knoxville, Tennessee, where the team grew from 3 agents to 22 and keeps 40 to 60 contracts moving at any given time. In the first episode of Keyed In, she walked through how her team decided which work AI should carry and which work stays with a person, and what it really took to get there.
The short version: they kept the human touch by deciding, one touchpoint at a time, what is routine enough for AI and what belongs to a person. Then they let AI handle the routine follow-ups and double-check their work, so people have more time for the calls and conversations clients remember.
Why the old system broke as the team grew
When Carrie joined the operations side three years ago, the team's transactions lived on dry erase boards that covered an entire office wall, with a square for every deadline and a check box for every document.
"When I came in three years ago, I was like, this is archaic. I was like, we cannot do this."
Carrie Hensley, SCH Real Estate
They moved to a spreadsheet-style tool, and it worked for a while. But SCH adds a lot of service to every file, so the checklist kept growing.
"We started off with like 25 tasks and we grew that to about 125 tasks per contract, because we are very much about the customer service."
Carrie Hensley, SCH Real Estate
By January the team had 18 agents, and it closed 38 contracts the December before, in what is usually a slow month. With that volume, an agent missed the end of an inspection period, final walkthroughs slipped, and "some customer service pieces were really falling through the cracks." The problem was no longer the number of tasks. It was that a growing team needed shared standards, and a static checklist couldn't keep everyone on them.
What a smooth transaction looks like behind the scenes
For SCH, a smooth transaction is not one without problems. It is one where the client is never surprised. When an agent submits a new contract, the operations team takes it from there:
- A phone call to congratulate the client, go over the disclosures they'll sign, and walk through their timeline: earnest money, inspections, closing, and whether they'll need a remote notary.
- An introductory email that recaps everything from the call.
- Outreach to the lender and title, including the buyer's lender when SCH is on the listing side, because "sometimes if you don't poke them, they don't always do what they're supposed to do."
"Our client knows about it. They know what's going on. They're not scared."
Carrie Hensley, SCH Real Estate
That is the bar every tool has to clear. If something breaks mid-contract, the client should hear about it from the team, early, with a plan.
The rule for any new tool: strengthen the relationship, don't replace it
Carrie was not quick to trust AI. She had heard that you have to teach it before it is useful, and that it can pull false information. What decided it for her was the team's motto, which the founder holds every agent to.
"Our big motto is no sales, just service."
Carrie Hensley, SCH Real Estate
So she wasn't looking for something that would "just send emails" or "just check the box." She needed a tool that would make the service better, and she met with ListedKit three or four times before buying to make sure it would. If you are weighing AI tools for your own team, that is a useful filter: ask what each one gives back to the client relationship, not only what it takes off your plate.
Which real estate emails can AI send on autopilot, and which need a person?
This is where SCH drew its clearest line. Three emails run on autopilot because they are short, factual, and the same on every file:
- The 5-day lender follow-up. Has the buyer committed to the loan and applied?
- The 15-day lender follow-up. Has the appraisal been ordered, and when is it due back?
- The 7-day clear-to-close email to title and lender. Where is the file, and are there any issues? If nobody answers, the team keeps following up.
"Those emails are very cut and dry, very simple emails, I think, are very autopilot. But most of the other ones are pretty personalized and tailored."
Carrie Hensley, SCH Real Estate
Everything else is personal. The team still calls every client on every new contract, buyers and sellers alike, and still goes line by line through every document a client signs. For written updates, Carrie uploaded the email templates the team has refined over eight years, and Ava writes from those templates, filling in the deal details and personal touches. Someone on the team reviews each email before it goes out.
"We still call our clients every time they get a new contract."
Carrie Hensley, SCH Real Estate
Over time she gave Ava a bit more room. The 10-day seller reminder (what not to pack, mowing the lawn, getting ready for the buyers) had always read as a little bossy, so she let Ava rework the wording, and it came back more professional. Karan Khanna, ListedKit's co-founder, pointed out why that matters: teams often expect AI to copy their old system exactly, even when that system is what was causing problems. Leaving room for it to improve the weak spots is part of adopting it well. If you want a starting point for your own templates, our transaction coordinator email scripts cover the common touchpoints, and AI vs. automation for transaction coordinators explains the difference between a fixed template and an assistant that adapts.
Can AI catch missing signatures before closing?
Carrie shared the file that made her want a second set of eyes. During a busy stretch, the team found that all four signatures on a set of documents belonged to the seller. The buyers had never signed.
"We were a week from closing and we had no buyer signatures."
Carrie Hensley, SCH Real Estate
Nobody caught it in time: not the operations team, not title, not the other agent. When you are moving fast, you see signatures and assume the document is done. They fixed it, but it changed how they check files. Now Carrie and the team's broker still review every file themselves, then pull the documents from Dotloop and have Ava check that every signature, name, and field is where it should be.
"We're human. We're going to make mistakes. We check and balance each other, my broker and I. Sometimes it's nice to have that third party as well."
Carrie Hensley, SCH Real Estate
The order matters. People check first and Ava double-checks them, which is how the team got comfortable using AI for compliance at all.
How do you get a skeptical team to trust AI?
Carrie didn't hand the new system to her TCs on day one. She ran the first two or three contracts through it herself, still with reservations, and only then let the team try it on real files. The turning point came at lunch one day, when one of her TCs told her he was already done with the three new contracts that had come in the night before.
"I just finished three new contracts literally within two hours."
A TC on Carrie's team, as she recalled it
He also kept notes on what wasn't right yet, and the operations team sat down with ListedKit to work through them. That pattern, test it yourself, then test it with your team on live files, gives you real evidence before a full rollout instead of a demo's best case.
How long does it take to set up AI for a real estate team, and is it worth it?
Carrie was candid about the cost. Early on, Ava's math was off, some email formatting was "a little wonky," and setup was tedious. She spent nights and weekends loading the team's email templates, task templates, and compliance templates.
"Ava has to learn you. You have to learn Ava."
Carrie Hensley, SCH Real Estate
The payoff, in her words: setting up a brand new contract used to take an hour to an hour and a half. Now her TCs complete one in about 30 minutes, including the emails, the client call, and the initial document check.
"For us, time is money... To me, it was worth the time that we got back now."
Carrie Hensley, SCH Real Estate
Karan compared it to hiring: a new assistant doesn't perform on day one either. The difference is what happens the next time your process changes. With rigid software you rebuild templates and rules by hand, while an AI-based system can adapt as the work changes.
Keeping agents ahead of deadlines
Carrie's favorite feature is a small one. Agents, especially new ones juggling several contracts, need reminders before inspection periods end and final walkthroughs come up. The team still texts them, and Ava also puts those deadlines on the team calendar and sends invites to the agents, because, as Carrie put it, agents "live and die by their calendar."
From the Q&A
Did you have to create rules for what AI can flag or prepare versus what a person approves? Yes. SCH set up separate compliance and task templates by side of the deal (buying or selling) and by financing type (VA, FHA, conventional, or cash), since a cash deal needs no lender emails and land may not need an inspection. It took some tweaking at the start to teach Ava which documents each contract needs.
How long have you been using Ava? The team started talking with ListedKit in March, ran demos through April and early May, and rolled it out fully at the end of May.
Key takeaways
- Sort every touchpoint into autopilot or personal. Start with short, factual follow-ups (lender check-ins, clear to close) as autopilot, and keep calls and explanations with people.
- Write your standard down first. Upload the templates your team already trusts so AI drafts in your voice, then review before anything goes out.
- Have people check first, and AI double-check. A second pass on signatures and required documents catches what tired eyes miss.
- Pilot on live files before a full rollout. Test it yourself, then with one or two team members, and track what still needs fixing.
- Budget real setup time. Templates and rules take hours up front, and the time back comes after.
Map your own team's human touch
The field guide under the video turns this conversation into a worksheet. It includes a copy-and-paste AI prompt that interviews you (or reviews your sent emails) to name your team's communication style, a simple pattern for making any touchpoint repeatable (the trigger, your standard, the deal context, and the human decision), and an example of what that looks like after an inspection. You don't need ListedKit to use it.
Want the full story of how SCH runs its operations? Read the SCH Real Estate customer story, or see the next episode of Keyed In on how directors of operations are using AI to help their teams handle more deals. If you are still comparing tools, our roundup of AI tools for real estate is a good place to start.