Provectus Reinvents Specialty Insurance Underwriting with Generative AI on AWS

Provectus helped an international specialty insurance company transform its underwriting process using generative AI hosted on AWS. The client faced lengthy manual processing of complex technical survey reports in underwriting decisions. Provectus developed a GenAI solution that uses Cohere’s Command Large Language Model on Amazon Bedrock to summarize and extract key information from 100-page reports in about 10 minutes. This accelerated underwriting policy decisions from days to minutes, increased underwriting capacity, reduced product costs, and grew annual recurring revenue.

Industry
Insurance
Published
July 2024

Reported outcomes

10 days

timeTime & speed

10 minutestime

Strategic outcomes

Speed & agilityAccelerated underwriting from days to minutesScale & capacityIncreased underwriting throughputCost efficiencyReduced insurance product costsNew business modelGrew annual recurring revenue

Primary read

Use case focus

Showing 3 of 3

  • 1Document Summarization
  • 2Risk Assessment
  • 3Underwriting Automation
  • Manual underwriting required subject matter experts to spend days reviewing lengthy and complex technical survey reports before making policy decisions.
  • The underwriting process was time-consuming and limited by the number of experts available to process reports.
  • Provectus created a generative AI-driven solution leveraging Cohere’s Command Large Language Model hosted on Amazon Bedrock within the AWS cloud.
  • The model was fine-tuned with in-context learning for specialized summarization and extraction of engineering technical survey reports in PDF format.
  • This automation enabled rapid summarization and key information extraction, reducing the report processing time to about 10 minutes per report.
  • Reduced underwriting decision time from about 10 days to less than 10 minutes.
  • Increased underwriting throughput with less human intervention.
  • Reduced insurance product costs and increased annual recurring revenue (ARR) for the client.
Architecture

The solution uses Cohere’s Command Large Language Model hosted on Amazon Bedrock. The model performs summarization and key data extraction from long technical survey reports. This accelerates policy decision-making in specialty insurance underwriting.

Implementation partners2
Sources & evidence1
Groundedness: 5/5

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Was this useful?

Similar cases