Client Stories
"OSC's AI software has revolutionized our insurance operations by providing real-time insights, improving underwriting decisions, and enhancing customer experience. We can now make data-driven decisions faster, and our customers benefit from more personalized services"
Chief Underwriting Officer at Top 10 Global Life Insurer
Client Stories
10X Business Performance
Take a look at how some of our clients increased their business performance through the use of OSC AI's products and services.
Wildfire Spread Modeling
One of California's largest P&C insurer had multiple $1B+ capital loss events due to Wildfires, raising insolvency risk.
Current wildfire models calculated risk at a Zip code level, with 24 hours of fire propagation calculation taking 96 hours, even using just basic slope, aspect and wind information.
We built computational fluid dynamic equations that run at exascale supercomputer speeds on standard cloud x86 nodes, giving individual house level risk scores, and calculating 24 hours of wildfire spread in just 62 minutes.
Mortgage Underwriting
A top five North American bank needed to keep pace in a suddenly very high growth residential mortgage market.
This required improving time to proficiency for new underwriters, automating underwriting for smaller loans, and allowing experienced underwriters to handle higher dollar cases.
We designed a custom decision algorithm with full explainability across factors as complex as 10 co-signers and 20,000 decision factors. We also developed a decision support UI for underwriters, and an executive dashboard.
IFRS17 ALM
A top 3 global insurance company faced the impossible combinatorial explosion problem of IFRS17 regulations.
Their current ALM calculation took 40 minutes each for 2,000 scenarios, i.e. 8 months. They ran their economic capital numbers just once a year and used Best Estimate Liability inter-quarter.
Our stochastic-in-stochastic distributed Monte Carlo algorithm reduced 8 months to 3 minutes. This allowed the client to project assets and liabilities for 60 forward years on a nested quarterly basis, looking at over 1 million scenarios.
Federal Reserve Stress Testing
Four global top 40 banks were faced with the same issue, CCAR compliance that is also an operationally effective asset.
Hand building models is a highly "artisanal", expensive, and error prone process that banks spend 8 months a year building for Fed stress tests, but these provide no insight into real time liquidity.
Our brute-force algorithms take the banks' line of business revenue, 20+ Fed mandated factors, 30+ market indicators and calculates 3000 transforms and exhaustively builds 10 billion models in software in minutes, accessible to CRO's via an intuitive UI.
Insurance Underwriting Automation
A top 10 US life insurer was losing market share to more agile competitors who had poorer risk management practices.
The current high quality underwriting process required an expensive, invasive, and time consuming blood draw, leading prospects to defer purchase. The insurer also incurred the cost for medical exams and alienated agents.
Our fluid vs fluidless triage and fluidless underwriting models allowed 45% of decisions to be automated with instant, fully explainable decisions, with the same risk levels as experienced human underwriters, and provided precise actuarial oversight.
Document AI for Asian Languages
A top 3 global reinsurer wanted to grow market share in Asia by providing clients with advanced digitization and underwriting.
Medical lab reports and application forms in Japanese and Chinese have a large character set. complex table layouts and data that may be a thrice faxed document with barely human readable data.
Our large vision model and advanced medical NLP model is the #1 for Asian languages, with 3 of top 5 global insurers adopting the platform. The product includes an Augmented Reality app for mobile capture, and a back office UI for data teams.
