I Built an AI That Calls My Real Estate Leads Instantly — Here's How
90% of real estate leads never get a call back. That’s not a number I made up — it’s a well-documented problem in the industry, and I was part of it. Not because I didn’t care about leads, but because my attention was on the clients already in my pipeline. Cold Facebook leads that came in at 2 p.m. on a Tuesday weren’t getting called that day.
Speed to lead is everything. The data on this is clear: the faster you respond to a new lead, the more likely they convert. But for a solo agent — or even a small team — responding immediately to every inbound lead is not realistic.
So I automated it. I built an AI voice agent that calls every Facebook lead the moment they submit my form. It qualifies them, handles objections naturally, checks my calendar for real-time availability, and books appointments. I get booked meetings — not leads I have to chase.
Here’s the full system.
What It Sounds Like
Before I get into the technical setup, let me show you what this actually sounds like. I ran a test call where I submitted my own form and played the role of an interested lead.
The agent opened: “Hi, I’m Abby from William’s team at EXP Realty. You signed up to receive a list of single-story homes in Austin from our downsizing ad. I wanted to follow up and make sure we send you the most helpful options. Do you have a quick minute?”
I said yes. The agent asked about my timeline — whether I was thinking about moving within the next three months. I said I was an empty nester who got my property tax bill and wanted to downsize. The agent acknowledged that specifically (“Saving on property taxes and downsizing as an empty nester are definitely great reasons to consider a single-story home”), then proposed a free 15-minute consultation.
When I asked about scheduling for the next day instead of the day of the call, the agent checked my actual calendar, came back with two available slots for the following day, and booked the meeting when I confirmed 11 a.m. worked.
That’s a real sales conversation. It includes context from the ad, a relevant qualification question, empathy in the response, a specific offer, and a real-time booking. The person on the other end had no reason to think they weren’t talking to a human assistant.
The Technical Stack
Facebook Lead Form — this is the trigger. The form collects name, email, and phone number. The field order matters: email first (easiest to give), then name, then phone number. You ease people into giving their phone number rather than leading with it. Facebook also auto-populates all three fields for most users, which helps completion rates.
Include a privacy policy disclosure on the form. This notifies leads that they may receive a call, which is both a legal requirement and a practical signal that they’ve opted in.
Zapier — the connection layer. When a Facebook lead is submitted, Zapier catches it and passes the lead information to 11 Labs. I use Zapier because it has a native Facebook Lead Ads trigger that works reliably. It also connects easily to Follow Up Boss, which handles my CRM side.
11 Labs — the voice AI platform. This is where the actual voice agent lives. 11 Labs offers some of the most natural-sounding AI voices available. The agent sounds like a real person. The platform handles the voice synthesis, the conversational AI, and the actual phone call — using a Twilio number.
The reason I chose 11 Labs over other platforms that also use 11 Labs under the hood: when 11 Labs is the call origin rather than a secondary layer, the latency drops significantly. The voice response times sound like a real conversation instead of a slightly delayed robot.
Tidy Cal — calendar integration. The agent checks my Tidy Cal calendar in real time to find available slots. I tell the agent in the system prompt to offer slots over the next 72 hours starting one hour from now — not immediately, so I’m not caught off-guard.
Follow Up Boss — CRM. Every lead that comes through, whether they book an appointment or not, gets added to Follow Up Boss and enters a nurture sequence.
The Agent’s System Prompt
The system prompt is where you train the agent’s personality, context, and behavior. This is the brain of the operation.
Mine tells the agent:
- Who they are (inside sales agent for a real estate team at EXP Realty in Austin)
- The context of the call (inbound lead from a downsizing ad for single-story homes)
- Tone (energetic, warm, professional)
- What the call is trying to accomplish (qualify timeline, book a 15-minute consultation)
- How to handle objections
The ad-to-script congruency matters. If someone responded to an ad about downsizing into a single-story home, the agent’s opening should reference that specifically. A generic “you submitted a real estate inquiry” opener kills trust. The opener I used — referencing the specific ad they responded to — signals that this isn’t a random cold call.
I also load the agent with contextual knowledge in the knowledge base: information about downsizing in Austin, what single-story homes are available in which neighborhoods, common objections from that buyer segment and how to address them. The more specific the knowledge base, the better the conversation quality.
What the Qualification Flow Looks Like
The two filter questions I have the agent ask:
- Are you thinking about moving within the next three months?
- What’s prompting your move?
The first question filters by timeline. Leads who are 12+ months out get tagged differently in my CRM and go into a longer nurture sequence. Leads who are within three months get pushed toward booking.
The second question is about gathering context, not filtering. Whatever reason they give — empty nesting, property taxes, job change, family situation — the agent acknowledges it specifically before moving to the offer. That acknowledgment is a basic sales technique: make people feel heard before you ask them to do something.
Leads who book get a calendar invite with William’s name on it. Leads who don’t get added to Follow Up Boss and receive follow-up communication.
Scaling This System
Right now this is set up for one ad targeting downsizing buyers. That’s the unit. One ad, one agent, one qualification script tuned to that specific buyer persona.
The structure scales horizontally. For every new ad I run — luxury buyers, investors, first-time buyers, relocation leads — I create a corresponding agent with its own system prompt and knowledge base. The ad and the agent match each other. The lead calls in from the downsizing ad, they get the downsizing agent. The luxury buyer fills out the luxury form, they get the luxury agent.
For team leaders and brokers, this creates a compelling pitch. Your agents don’t prospect — they show up to appointments that the system already booked. That value proposition changes how you recruit.
What I’m Building Next
The current system is the inbound agent — it handles the first call when a lead submits a form. The next layer is a follow-up agent. That agent would review the call transcript from the initial conversation, pull in any other information the lead provided, and generate the right follow-up sequence — the right timing, the right message, the right offer based on where they are in the funnel.
Right now the two are separate. Eventually they’ll be connected: the inbound agent hands off to the follow-up agent, and the follow-up agent continues working the lead until they’re ready to move.
The AI follow-up starter kit — which walks through the exact setup I described above — is available as a free download (linked in the video above). It gives you the step-by-step process in PDF format so you can work through it yourself.
For a broader look at the AI tools I’m using across lead generation, follow-up, and content, the tools page has the full picture. And if you want to stay current as I build out the next version of this system, subscribe to the newsletter.
Liked this article? Get more like it.
AI tools, prompts, and workflows that close deals — delivered in 5 minutes a week. Free, unsubscribe anytime.