Business Impact

  • ~85%

    Classification Accuracy

  • ~70%

    Extraction Accuracy

  • 5

    Types of Document Classified and 13 Fields Extracted

Customer Key Facts

  • Industry : Insurance
  • Country : United States
  • Size : 30000+

Problem Context

The customer is one of the largest commercial insurers (property and casualty) in the world. They wanted to classify Submission Documents, extract required and relevant Information and enhance extracted data using third-party APIs.


  • Extensive time spent by underwriters on a low value, repetitive work
  • Low quality and transparency of broker quotes
  • High error rate due to manual processing

Technologies Used

Pub/ Sub
Cloud Function
Vertex AI
Cloud SQL
API Gateway
Cloud Scheduler
Cloud Storage
Virtual Machine


Quantiphi re-engineered the submission intake process from ingestion, classification, and extraction to data enrichment with third-party APIs.

  • Email ingestion: Real-time email ingestion system for ingesting submission documents
  • Document Classification: Classify document to determine document type and content
  • Extraction: Extract multiple fields, values, and embedded objects from emails and submission documents
  • Data Enhancement: Leverage Dun & Bradstreet API to correlate and enhance extracted information
  • Automated Submission Triage: Application triage based on submission and its size, complexity, horizon, and application completeness for scoring/routing


  • Industry-leading classification and document extraction accuracy
  • Streamlining and standardizing underwriting workflows
  • Dashboard for increased visibility and resolution speed

Looking for similar project?

Let's Talk

Get your digital transformation started

Let's Talk