Back to Projects

AI Voice Receptionist for Real Estate

A Vapi-based voice agent handling inbound buyer calls — qualifies budget, timeline, and area, books showings, and hands off to agent for high-fit leads.

2026 AI Voice Engineer AI & Bots

VapiOpenAITwilioGoHighLevelCalendar BookingCall Recording

taimoorakhtar.com/projects/ai-voice-receptionist

Introduction

A real estate brokerage was losing leads after-hours and on weekends. The receptionist was overwhelmed Monday mornings dealing with voicemails from Saturday's open houses. Most buyers wanted to talk to someone immediately — and a polite voice-AI that could actually qualify and book was a measurable upgrade over voicemail.

The Challenge

Voice AI in 2025 still had a robotic-cadence problem. The first deployments I tried sounded like Siri reading a teleprompter. Buyers hung up within 20 seconds. The challenge was making the agent sound human enough to hold the conversation through full qualification, while staying inside compliance guardrails for real estate (no price guarantees, no legal advice, no fair-housing missteps).

The Solution

Deployed on Vapi with a tuned voice model, conversational prompt enforcing strict qualification flow, and a Twilio number for inbound routing. Bot handles qualification (BANT-style adapted for real estate), books showings against agent calendars, and escalates high-fit calls to the on-call agent in real time.

Technical Deep Dive

Voice model tuning. Vapi voice model selected for naturalness and Pakistani-buyer-friendly accent comprehension. Disfluencies ('um', 'let me check') intentionally included to humanize. Latency tuned below 800ms for natural turn-taking.

Qualification flow. Bot collects: budget range, timeline (immediate / 30-day / 90-day / exploring), preferred area, financing status, current home situation. Stored as custom fields on the GHL contact created from the call.

Live calendar booking. Bot reads agent availability from GHL calendar via API, offers 2-3 slots in the prospect's stated preferred area. Books, sends SMS confirmation, calendar invite to agent.

Compliance guardrails. Prompt explicitly forbids: quoting specific prices, making legal claims, discussing protected-class topics. Edge cases trigger immediate handoff with a 'let me transfer you to the agent' line.

Real-time handoff for high-fit. Calls scoring above qualification threshold (immediate timeline + matching budget + verified financing) get transferred live to the on-call agent's mobile via Twilio warm-transfer.

Key Features

85% inbound call answer rate (up from ~40% with human-only)
Full BANT-style qualification in voice
Live calendar booking during the call
After-hours coverage with no voicemail backlog
Real-time warm transfer for high-fit prospects

Results & Impact

12-15 qualified contacts per day from previously-lost after-hours volume
Average qualification call length: 4 min 20 sec
Showing booking rate from inbound calls: 38%
Zero fair-housing compliance incidents across 4 months

Lessons Learned

"Voice AI works when you constrain it tightly. Open-ended LLM voice is a liability — narrow scope is your friend."

"Compliance guardrails in real estate are not optional. Build them into the prompt with explicit forbidden topics."

"The handoff threshold matters more than the qualification depth. Get high-fit calls to a human fast."

Related Work

Have a similar build in mind?

I'm available for engagement on GoHighLevel implementations, A2P 10DLC compliance, AI automation pipelines, and CRM migrations. Most projects start at $300–$1,200 depending on scope.

Book a 30-min call WhatsApp View More Projects