legal_document_review · finance · workflow
Accelerating insurance policy reviews with generative AI: Verisk's Mozart companion
Verisk's insurance customers spend significant time regularly reviewing changes to policy forms, with change adoption taking days or weeks due to the complexity and unstructured nature of legal policy documents.
How it works
Common implementation structure
How this type of workflow is generally built, generalized across documented cases — not tied to any one vendor's stack. Click any stage to read what happens there. Specific products that implement these stages appear in “Tools commonly seen” below.
Stage 1 · Periodic document ingestion and embedding
An AWS Batch job periodically reads policy documents from Amazon S3, chunks them into smaller slices, creates embeddings via Amazon Titan Text Embeddings through Amazon Bedrock, and stores them in an Amazon OpenSearch Service vector database.
Tools used
Amazon BedrockAmazon S3AWS BatchAmazon Titan Text EmbeddingsAmazon OpenSearch ServiceAWS LambdaClaude 3 SonnetLangChain
Outcome
The Mozart companion generates over 90% good or acceptable summaries and reduces policy change adoption time from days or weeks to minutes, increasing productivity and enabling timely implementation of changes.
What failed first
The initial generative AI model results were not close to the desired level of accuracy and consistency, requiring iterative redesign, multiple foundation model calls, and testing of various foundation models before achieving acceptable quality.
Results
Time savedfrom days or weeks to minutes
Volumeover 90% good or acceptable summaries
Grounding & classification
Source type: technical build writeup
31 fields verified against source quotes.
data extractiondocument airagsummarizationcontractpolicy documentfailure mode describedhuman review describedmetric backednamed customertools describedvendor confirmedworkflow describedinsuranceaccuracy improvementcycle time reductionemployee productivitytechnical build writeupcompliance monitoringlegal document reviewdocument to recordrag answering