Frictionless Underwriting - Commercial Insurance Risk Assessment powered by GenAI

Insurance Underwriting is a very complex, manual, and time-consuming process. Underwriters typically must skim through large contract documents and Insurance proposals manually. Risk assessment is done manually, making the process tedious and slower. Hence, their ability to process or reject the application slows down, resulting in a lower quote-to-bind ratio.

Innovations in Generative AI are transforming organization’s ability to make quicker and accurate decisions. Risk assessment process in the Insurance industry is getting transformed with the advent of Generative AI.

Our solution leverages our AI accelerator, the Intelligent Document Processing (IDP) framework (on Azure) to extract relevant contents and risks from documents. Generative AI then helps categorize and summarize the various risk types, generate risk scores across all the risk types and finally allow easy querying options to the underwriter.

8-Week PoC Summary:

We will conduct 8-week PoC which will involve data quality and relevancy assessment, feasibility analysis, model selection, training & integration requirements, User Acceptance, scalability assessment,

The output will involve building an MVP / prototype for business and technical validation.

Research indicates that Underwriters spend more than half their time in non-value adding activities, rather than evaluating proposals and collaborating with brokers & agents to secure business growth.

Eviden Solution:

This Risk Assessment Solution will help underwriters to capture and score risks accurately and use the information to make decisions. This solution will help transform the risk assessment process and thereby improve quote-to-bind ratio.

Our solution leverages our AI accelerator, the Intelligent Document Processing (IDP) framework (on Azure) to extract relevant contents and risks from documents. Generative AI then helps categorize and summarize the various risk types, generate risk scores across all the risk types and finally allow easy querying options to the underwriter.

Content Extraction: Unstructured data from the broker slip document is extracted using Natural Language Processing (NLP)
Document Summarization: The solution leverages the Generative AI capability to provide a summarization feature, that captures a high-level summary of the contents present in the contract document
Risk Categorization: The IDP and Generative AI engine extracts and categorizes all risks associated with the respective contract document and provides a summary at individual risk level
Risk Scoring: Machine Learning algorithms provide a risk score against each risk type. The score is based on probability and severity of risk occurrences, scored on a scale of 0 to 1
Risk Assessment Co-Pilot: Standard and free text querying options allow underwriters to generate responses to their queries related to the contents in the document.

8-Wk PoC phase will involve the following activities:

Problem / Goal Definition: Clearly articulate the business goal by defining scope and boundaries of PoC/MVP.
Feasibility Assessment: Consider factors such as data availability and quality, process complexity and capabilities of the Gen AI models in relation to the same
Model Training requirements: Evaluate use of pre-trained / custom models.
Integration with existing systems: Assess how seamlessly Gen AI can be integrated in current systems and workflows, ensuring compatibility with existing technologies and evaluate potential modifications that may be required.
User Acceptance – Gauge end-user acceptance on MVP. Solicit feedback and refine models in tune with client expectations. Identify potential risks, including technical challenges, unforeseen biases, and any adverse effect on stakeholders.
Create production level architecture for scalable solution.

Azure Set-up:

We can help the customer with the right solution architecture for Gen AI and document processing solution using Azure.
We offer Professional service to architect the solution using best practices of Microsoft Azure and also support in managing services on Microsoft Azure
A dedicated Azure subscription will be required in customer sandbox / non-prod environment to enable PoC development and testing

Adobe Support and Services

Adobe US

Détails

Liens