Empowering a Leading Healthcare Provider Boost Customer Satisfaction by 60% with GenAI
About the Client
The client is a leading digital health company specializing in AI-powered virtual assistants for healthcare systems. They empower over 1.5 million monthly active users to navigate the complexities of their healthcare journeys, partnering with Integrated Delivery Networks (IDNs) across the US, and deploying their technology on websites, mobile apps, and patient portals.
Their Challenges
Navigating the complexities of insurance policies and procedures can frustrate patients and healthcare providers. The client identified several key challenges hindering their ability to deliver exceptional customer service:
Limited Adaptability of Existing Chatbot System
Their current chatbot system, built on a leading cloud provider, needed more flexibility to handle the nuances of evolving health insurance inquiries. This resulted in difficulty keeping pace with the ever-changing complexities of healthcare plans.
Inefficient Information Retrieval from PDFs
Health insurance policies are often dense and complex documents stored in PDF format. The existing system struggled to extract and utilize this critical information efficiently, hindering its ability to answer specific policy-related questions.
Fragmented User Experience for Providers and Payers
The current solution catered to two distinct user groups – hospitals (providers) and insurance agencies (payers). This resulted in a less-than-ideal experience for both parties, hindering overall customer satisfaction.
The Solution
To address these challenges, the client partnered with Searce to develop a groundbreaking solution leveraging Generative AI and Google Cloud’s suite of AI services. This innovative system tackles the complexities of information retrieval and comprehension, specifically designed for the healthcare industry.
Here’s how the solution breaks down:
Structured Data Extraction with Searchable Knowledge Base
The system utilizes a technique called Structured OCR to meticulously extract text from PDFs while preserving the original structure and layout. This extracted data is then transformed into a searchable format, creating a comprehensive knowledge base of health insurance policy information.
IMPACT
Enhanced Accuracy
Trained on health insurance PDFs, the LLM now provides insightful answers to complex questions
IMPACT
Improved Customer Service
Hospitals & agencies can offer a more efficient self-service experience, boosting customer satisfaction
Semantic Understanding with Embeddings
Vertex Embeddings, a Google Cloud offering, plays a crucial role. It converts the processed text from PDFs into a unique numerical representation, essentially a “fingerprint” for each piece of information. This allows the system to understand the meaning and context of the extracted data.
Efficient Information Retrieval with Langchain and Vector Search
Since large language models (LLMs) can’t handle massive amounts of data at once, the system breaks the information down into manageable chunks using Langchain. These chunks are then fed into a Vertex AI Vector Search engine, which acts like a super-powered index. This engine efficiently finds the most relevant information based on a user’s question within the knowledge base.
IMPACT
Streamlined Operations
By automating routine Q&A tasks, the client can free up valuable resources for more strategic initiatives
Success Story Highlights
Searce’s solution for the client boosted customer satisfaction by 60% and slashed resolution times by a staggering 72%. Structured OCR preserves 95% of the information’s structure, while Vertex Embeddings process the data. This empowers the virtual assistant to deliver insightful answers within milliseconds (<100ms latency), resulting in more streamlined experiences and freeing up resources.