•  Customer is a large property management firm managing properties in the US
•  As part of their property sale process, they used to receive questionnaires from lenders (like banks, financial institutions, mortgage firms etc.) and property attorneys regarding the ‘in sale’ property
•  Pretty much the questions will be repetitive and hence the responses too. It was identified as a right candidate for automation using AI/ML

Introduction

Customer is a large property management firm managing properties in the US

Challenges

•  A comprehensive study of the questionnaires and responses was done. Questionnaires were identified to be from banks, attorneys and generic ones including PDF, Word and Text formats. Documents included scanned and regular contents
•  LLM (Large Language Model) based algorithm was identified to be the right solution since questionnaires and responses included unstructured data and natural language

Solution

LLM based solution was built which had following process steps:

•  Upload of a new questionnaire into a folder
•  Processing of the questionnaire by an AI engine
•  Output response was provided in the form of an Excel/Word file
•  User Interface was offered to make corrections/revisions as appropriate
•  Feedback loop was constructed to process the revised data

Key Benefits

•  Drastic reduction of processing time from number of days to a few minutes thereby increasing their business process efficiency
• Reduced error rates as responses were based on past data and no subjective creation of responses
• Feedback loop in the solution ensured that the corrections/revisions made by the operators were incorporated further thereby enhancing the accuracy on an ongoing basis