works

VOICE AI AGENT

DCM (Dot Com Media), a client acquisition and conversion company, approached us with the need for an intelligent inbound callback agent to replace their sales development representative role. Their goal was to automate the initial callback conversation with law firm staff and business owners, qualify leads efficiently, and book high-quality Zoom appointments — all while maintaining a natural, human-like conversation experience.

Project Overview:

We partnered with DCM to design and deploy a state-based conversational voice AI agent named Samantha, built on the Retell AI platform. The objective was to create an agent that could handle inbound callbacks from attorneys, paralegals, and law firm staff, guide them through a qualification flow, handle objections persistently, and book diagnostic Zoom appointments — without sounding scripted or robotic.

‍

Objectives:

The primary objective was to replace the manual SDR callback process with an AI agent capable of qualifying leads and booking appointments at scale. We aimed to build a natural, confident conversational flow that mirrors how a real sales development rep would handle these calls. The agent needed to handle objections up to three times using different angles, route gatekeepers appropriately, collect booking information with precision, and improve appointment show rates through structured commitment steps

Scope of Work:

Our work on this project spanned across both design and development phases:

• Conversation Architecture & Prompt Engineering: Designed a state-based conversational agent on Retell AI with structured flows for opening, qualification, objection handling, and booking — built to sound natural while following strict sales logic.
• Objection Handling System: Built a multi-category objection framework with up to 3 persistence attempts per objection, each using a different angle, ensuring the agent never gives up too early while avoiding aggressive or repetitive responses.
• Booking & Availability Integration: Integrated live tool calls for real-time calendar availability checks and appointment booking, with strict scheduling logic, timezone handling, and controlled time slot offers.
• Information Collection Flow: Implemented a precise multi-step data collection sequence covering email, full name, business name, website, and phone confirmation — with phonetic readback rules and hard stop logic to ensure accuracy.
• Testing & Optimization: Conducted iterative QA across real call transcripts, refining prompt logic, post-call variable behavior, and objection handling based on observed agent performance in live calls.