MRH Trowe provides insurance, finance, and risk management services, helping businesses and individuals protect assets, mitigate risks, and achieve financial goals.
Challenge
MRH Trowe’s account managers manually reviewed lengthy, unstructured PDF property reports to extract required “green” and “yellow” field data into Excel for insurer review. This process took up to 2.5 hours per document, was prone to errors, and limited capacity for client relationship management.
The company needed an automated solution to process multiple documents quickly, accurately, and consistently, while still allowing human oversight before submission to insurers.
Solution
Firemind developed an AWS-based automated workflow using generative AI to extract and organise property data. PDF files were processed through Amazon Textract for text extraction, AWS Lambda for data preprocessing, and an LLM hosted on Amazon SageMaker for structured data inference.
AWS Step Functions orchestrated the process, and the extracted data was output into CSV format for rapid review and submission. This reduced document review times from hours to minutes, enabling greater efficiency and freeing account managers for higher-value client work.
Services used
- Amazon Bedrock
- Firemind’s Pulse
- Amazon SageMaker
- Amazon S3
The Results
- Reduced extraction time from 2.5 hours to 7 minutes
- Delivered in just 4 months from first meeting to sign-off
- Higher accuracy in both “green” and “yellow” field data
- Consistent, repeatable extraction process
- Increased account manager availability for client engagement
