AI in Insurance: Resolution Life's AI Journey with Rajesh Malla
- Observability
- H2O Driverless AI
- H2O Hydrogen Torch
Overview
Rajesh Malla of Resolution Life, one of Australasia's largest life insurers, walks through how the company used H2O Driverless AI to build a claims triage model. The talk covers the business problem of slow, manual claim segmentation, how the model now classifies a claim within 15 seconds of submission, and the measured jump in accuracy and cost savings over the in-house model it replaced. He also outlines where the team plans to extend H2O next, from call center reason prediction to a product knowledge library.
Key takeaways
- Why claims triage was a bottleneck: manual segmentation that could take a case manager a day to a week per claim
- How Resolution Life deployed an H2O Driverless AI Mojo model on Snowflake, running on Azure, to segment claims within 15 seconds of submission
- The measured accuracy gain: from 71% with the prior in-house model to 77% with H2O, built in roughly a quarter of the development time
- The estimated financial impact, described as roughly double the value of the in-house model it replaced
- Where the team is expanding next: predicting call center contact reasons, and using Hydrogen Torch to build a product disclosure statement knowledge library
- Why model explainability and interpretability mattered for internal governance and audit purposes
Start owning your AI