Introduction to Client and Industry Context

We worked with India’s leading insurance aggregator, which operates with a large network of sales agents responsible for selling policies to customers over the phone or in person. However, when prospective customers inquired about specific policy features, agents faced delays due to the need to send queries to their team leaders for clarification. This created a bottleneck in the sales process.

These inefficiencies negatively impacted both customer experience and agent productivity, as valuable time was wasted on back-and-forth communication.

To address this, the company sought our help in creating an AI-powered assistant that could provide agents with accurate, contextually relevant answers to policy queries in real time. Their goal was to improve productivity and streamline the sales process without sacrificing service quality.
‍

Problem Statement

Operational Bottlenecks:
‍Agents were dependent on team leaders to get answers to policy-related queries, which caused delays and hampered productivity.

Customer Experience Gaps:
‍Prospective customers, especially high-value leads, were left waiting for responses, affecting their experience and the company’s ability to scale effectively.

Financial Impacts:
This productivity loss across a large network of agents negatively impacted the company financially, stalling business growth and profitability.

How the Insurer Was Solving the Problem

Initially, the company implemented a question-answering system based on the RAG (Retrieval-Augmented Generation) framework, which allowed agents to retrieve answers from policy brochures quickly. They integrated a chatbot interface to make it easier for agents to query the system and get responses. However, the system still faced several challenges:

• Response Time: The system took up to 30 seconds to provide answers, which was too slow for agents dealing with live customers.
• Downtime: The system often had technical issues, causing periods of unavailability, which hindered agent productivity.
• Limited Data Coverage: Only 8 plans were available initially, forcing agents to manually search for information on other policies.

Solution Design and Rationale

We were brought in to optimize the solution and address the challenges the insurer faced. Our solution included the following:

• Improved Data Pipeline: We designed a more efficient and robust data injection pipeline to enhance the speed and accuracy of data retrieval.
• Revamped Database: We upgraded the database and introduced new methods for data storage and retrieval, improving system efficiency and scalability.
• Expanded Policy Coverage: We updated the QnA system to support queries across 25+ insurance plans from 8 different insurers.
• Open-Source LLMs: We incorporated open-source Large Language Models (LLMs), which significantly reduced latency and improved the system’s response time, making it faster and more reliable.

How the Solution Worked

‍We integrated the company’s policy brochures into the RAG framework, linking it with GPT-3.5 (LLM). Here’s how it operated:

• Data Collection: We fed the policy brochures and detailed information from all insurers into the system.
• Chatbot Interface: Agents were provided with a user-friendly chatbot interface, enabling them to quickly query policy details and receive real-time responses.
• Instant Query Resolution: When an agent received a customer query, they simply typed the question into the chatbot, which would pull the most relevant information and provide an accurate response within seconds.

Business Impact and Quantifiable Results

Increased Reach and Engagement:

• Response times were reduced to under 5 seconds, enabling agents to engage with more prospects and increase their reach.
• Agents were able to handle more inquiries simultaneously, resulting in improved productivity.

Agent Efficiency:

• Agents spent less time seeking clarifications and more time selling, which boosted sales conversions and enhanced customer experience.
• The AI assistant allowed agents to focus on higher-value interactions, greatly increasing their overall efficiency.

Conversion and Success Rates:

• By providing faster, more accurate answers, the AI assistant helped agents improve customer decision-making, leading to higher conversion rates.

Cost Savings:

• The automation of routine tasks significantly reduced operational costs while also boosting agent productivity, allowing the company to scale without increasing overhead.