A leading commercial insurer relied heavily on manual review of broker submissions for complex RFQs. Each submission required specialist expertise, making it difficult to review every request in detail. The process was time-consuming, inconsistent, and limited by human capacity.
Key challenges
Fewer RFQs annually, with too little data for traditional automation methods.
The manual review process depended on specialist knowledge, limiting scalability.
Significant variation in broker submission phrasing and structure hindered standardization.
The solution
LLM-powered submission review
Validated and refined outputs.
Deployed LLM ingestion framework on the cloud.
Provided an interactive tool for testing and feedback.
Configurable and intelligent framework
Used RAG for fast large-document ingestion.
Enabled quick setup for new fields or formats.
Applied confidence scoring to flag uncertain results.
1
Collaborative development
Co-created solution with SMEs through iterative reviews.
Embedded feedback loops for continuous refinement.
2
Seamless deployment
Integrated within the cloud environment.
Ensured strong data security and compliance.
3
Continuous optimization
Monitored accuracy and performance metrics.
Updated configurations as submission formats evolved.
Business outcomes
High accuracy on unseen documents, cutting review time.
More small deals reviewed using freed-up capacity.
Maintained premium levels with notable COR gains.
Model performance
Tuned LLMs managed variation with limited samples.
Delivered consistent, high-quality extractions.
Scaled efficiently without losing accuracy.
Rapid improvement
Improved accuracy in a week.
Clear traceability of sources and prompts.
Continuous feedback accelerated optimization.
Transform your enterprise with AI that delivers
