AI Voice Agent for a Nationwide Fencing Company

An AI voice agent that answers, qualifies, schedules, and escalates inbound calls, 24/7 across all U.S. time zones.

What We Delivered

• AI-powered inbound call handling built on Retell AI
• Real-time appointment booking with rule enforcement
• Emergency call identification and escalation
• SMS & email confirmations and reminders
• Secure admin dashboard with full visibility

Impact

• 80% call pickup rate
• 30% increase in booked appointments
• 2–3 hours/day saved for office staff

Quick Facts

• Industry: Residential & Commercial Fencing
• Platform: AI Voice + Web Admin
• Scope: Nationwide (USA)
• Services: AI workflows, backend systems, admin dashboard, automation

Background

The client is a nationwide fencing company serving homeowners, property managers, and contractors across the United States. With a mix of residential (wood, vinyl, aluminum) and commercial/industrial projects, inbound call volume is constant—and time-sensitive.

Despite using multiple tools, the company relied on salesmen answering phones or voicemail. As volume grew, missed calls, slow responses, and lost emergency jobs became a recurring revenue leak.

The Challenge

Business

• Missed inbound calls during peak hours and after-hours
• Lost emergency opportunities requiring immediate response
• Office staff overloaded with repetitive call handling

Operational

• Inconsistent appointment booking
• No standardized intake for estimates
• Manual follow-ups when information was incomplete

Technical

• Need to operate reliably across all U.S. time zones
• Enforce strict scheduling rules without human intervention
• Provide leadership with visibility into AI-handled activity

The Solution

Stackup Solutions built an intelligent voice agent and admin system designed to own the inbound call workflow end-to-end.

1) AI Call Handling (Voice)

• Natural, conversational AI answers inbound calls 24/7
• Supports residential, commercial, and contractor inquiries
• Intent detection distinguishes regular vs emergency calls
• AI + scripted fallbacks ensure reliability
• Escalates to a secondary phone number when human intervention is required

2) Emergency Call Identification

• Emergency intent detected through prompting and training
• Priority handling ensures urgent jobs are never lost to voicemail
• Emergency appointments bypass standard friction where required

3) Real-Time Appointment Management

• Live scheduling during the call
• Rule engine enforced by AI:
  • 2 available slots per day
  • Multiple appointments per slot
  • Max 30 minutes per appointment
  • Prevents double-booking
• Politely reroutes callers when requested times violate rules
• Seamless voice-based cancellations with automatic locking

4) Estimation Intake

• AI collects structured estimate data during calls
• Captures service requirements and special instructions
• Logs all information for sales follow-up
• If information is incomplete, admins are notified to follow up manually

5) Notifications & Reminders

• Email and SMS confirmations via SendGrid
• Notifications sent to:
  • Client
  • Assigned salesman (round-robin assignment)
• Automated reminders:
  • One day before
  • Same-day morning
  • 15 minutes prior
• Reliable delivery using background job processing (Node cron + polling)

6) Admin Panel (Internal Tooling)

• Secure, web-based admin dashboard
• Built for daily operational use
• Features:
  • Calendar and list views of appointments
  • Detailed appointment records (status, type, emergency flag)
  • Call logs and AI-captured metadata
  • Visibility into routine vs emergency calls
• Enables office managers to monitor, validate, and follow up without listening to every call

Technical Architecture

• Frontend: React.js + Tailwind CSS
• Backend: Node.js + Express
• Database: PostgreSQL
• Infrastructure: AWS
• AI Voice: Retell AI
• Messaging: SendGrid (Email + SMS)
• Automation: Node cron + polling queues

The system was designed to be scalable, allowing future expansion into additional call flows, rules, or regions without re-architecture.

Implementation Process

• Discovery & Call Flow Design: Mapped real-world call scenarios including emergencies and after-hours traffic
• AI Prompting & Training: Tuned intent detection and fallback paths
• Scheduling & Rule Engine: Encoded appointment constraints directly into AI logic
• Admin Tooling: Built internal dashboard for trust and visibility
• Nationwide Timezone Handling: Accurate scheduling across U.S. time zones

Results

• 80% call pickup rate, eliminating voicemail dependency
• 30% increase in booked appointments
• 2–3 hours/day saved for office staff
• Emergency calls captured instead of missed
• Consistent intake and follow-up across all regions