G2 Logo UserLoveG2_V1
Search
Get Started
G2 Logo UserLoveG2_V1
IO_Cloud

Client : Insurance & Investment Services Company

Industry: BFSI
Location: USA

Leading enterprise event automation platform helping marketers & event professionals deliver measurable results from events.

partner_databricks-1

 
“ As a leading organization in the BFSI sector, we constantly strive to enhance our customer service and operational efficiency. Partnering with InfoObjects has been a transformative step in our journey. We are immensely satisfied with the professionalism and customized support provided by InfoObjects. Their Generative AI expertise and the technical excellence in executing this project were exceptional.“
2017-Logo-square-on-grey

Director of Engineering,
BFSI Fortune 100 Institution

Overview

InfoObjects has provided a strategic solution to a prominent BFSI client, addressing a critical challenge associated with prolonged case handling times. The primary issue stemmed from the manual labor involved in locating and retrieving pertinent documents necessary for resolving each case file, which significantly extended the resolution timeframe and frequently resulted in case escalations. To mitigate this issue, InfoObjects proposed the implementation of an advanced support tool. This innovative tool enables agents to efficiently access the required information through a simple chat interface. By leveraging a sophisticated Large Language Model (LLM) at its core, the tool ensures the delivery of precise and relevant information to the agents, thereby streamlining the case resolution process.

78 %

Reduced Manual Effort

68 %

Faster Case Resolution

71 %

Faster Data Retrieval

Our GenAi experts designed and implemented a RAG ( retrieval Augmented Generation) based LLM pipeline to collect disparate documents (webpages, internal wiki,media), then store them as Embedding in vector database, which can be later retrieved by a similarity search and feed to LLM to obtain the relevant data

 

Approach

Data Preprocessing

  • Built a data pipeline to support historical and incremental loads
  • Convert disparate source information into standardized format
  • Chunking Strategy

Governance

  • Risk and Guardrails
  • Multiple Feedback mechanism to Evaluate and improve performance

LLM Integration

  • Vector Embedding
  • Search and Retreival
  • Prompt Engineering
  • Integrated UI

Technology Stack

  • Azure OpenAI
  • Databricks
  • GPT 3.5 Turbo

CONCLUSION

InfoObjects implementation of RAG based tool helped enterprise to get immediate answers accessing data stored across myriad documents and internal web sources . Currently , users can simple “chat” with the tool which provides immediate insights on live data using LLM. Our GenAi experts helped the client to reduce the time to retrieve and learn from data exponentially .