Skip to content

AI Engineering · India

PolicyBoss - A Voice AI Agent for Motor Insurance Policy Verification

A voice AI agent that supports policy verification calls alongside live agents, retries unfinished verifications, and follows up on WhatsApp after every attempt.

Discuss Your AI Project
  • Real-Time Voice AI
  • Live Agent Assistance
  • Scheduled Call Retries
  • WhatsApp Follow-Ups

About the client

PolicyBoss’s verification team confirms motor insurance policy details with customers over the phone. It combines AI-assisted calling with live agents, so an agent can update requirements, answer questions and resolve issues during the call. The system works in multiple languages and sends automated reports over WhatsApp. The goal is a faster, more accurate verification process with less operational effort.

Industry
Insurance / InsurTech
App type
Multichannel Voice AI Agent for Policy Verification
Practice
AI Engineering (Build)
Services
Agents & Agentic Workflows, AI Infrastructure & LLMOps, AI Quality Engineering & Evals, AI Platform Partnerships & Professional Services, AI Strategy, Governance & Compliance Advisory

Tech stack

  • Next.js (frontend)
  • FastAPI (backend)
  • PostgreSQL (database)
  • Google ADK (AI call assistance)
  • WebSocket
  • GCP (cloud)
  • Exotel (telephony)
  • WhatsApp API (post-call communication and reporting)

Engagement summary

We built a voice AI agent that supports PolicyBoss’s verification calls. Exotel starts each call and connects it to our AI and agent service. Google ADK and WebSockets keep the conversation running in real time with low delay.

Missing or updated policy details are recorded as they come up in the call. After every call attempt, a WhatsApp message and a report go out automatically.

Calls that don’t complete are retried up to 6 times across 3 days, with 2 calls per day. Once a verification succeeds, it’s logged and synced to the CRM.

Challenges faced by the client

PolicyBoss’s verification process had gaps on both the technical and the process side.

  • Real-time audio

    Streaming and sampling live call audio through Exotel was difficult to integrate.

  • Follow-ups and retries

    Follow-ups after missed or incomplete calls weren’t seamless, and the retry process in place was inefficient and led to drop-offs in verification.

  • Manual updates

    Updating verification requirements was manual, which caused errors and delays.

  • Fragmented communication

    Customer communication was spread across voice, WhatsApp and reports.

Results

  • Unfinished verifications get scheduled retries and follow-ups

  • Questions on policy terms, coverage and status are answered during the call

  • Missing or incorrect policy details are updated during the call itself

  • Voice, WhatsApp and reports work together as one system

  • Verification is more compliant and transparent

How it works

  • The agent holds audio-to-audio conversations in multiple languages.
  • Live agents work alongside the AI on the same call.
  • Completed verifications are logged and synced to the CRM.

Core solution

Call initiation

Exotel triggers the verification call and connects it to our AI and agent service.

Capturing requirements, resolving queries and sending confirmations all happen in one system, integrated with Exotel telephony and the backend systems.

More Case Studies

View All Case Studies

Let’s Explore How We Can Build Together

Whether you’re building a new product or modernizing an existing one, get insights from our product, engineering, and AI experts.