Privacy-Aware AI for Insurance Decision-Making

Interpretable, Causal AI

Checklist

Fully
Compliant

  • No filings or amendments required
  • No Personally Identifiable Information used
  • Statistical parity on protected classes
  • SOC2 compliant
Medal

Proven
Success

  • Trusted by top 10 P&C and L&H Carriers
  • Double-digit improvement across functions and business lines
  • Complementary to existing models
Research driven

Research
Driven

  • Powered by published academic research in top scientific journals
  • 100% interpretable findings
  • Causal graph for deep insights

Selected Use Cases

Conversion

200%+ Conversion Improvement

Waffle provides early-funnel conversion indicators across personal lines without needing personally identifiable information. This innovation revolutionizes customer engagement and agent follow-up.

Case Study: Over 200% conversion improvement for one of the largest US L&H Carriers.

Loss ratio

10%+
Loss Ratio Improvement

Waffle develops precise, early loss-ratio models using proprietary leading indicators and advanced modeling techniques.

Case Study: 10+% improvement in loss ratio for a Top 10 US P&C Carrier.

fraud detection

Efficient Fraud Detection in < 1 Second

Waffle’s AI models enable efficient, real-time causal factor discovery, identifying new factor ratings, potential fraud, and improvement targets.

Case Study: Double-digit improvement in fraud detection for a Top 10 Global Carrier's Small Business Insurance.

Access Options

Base Model

The base model is available for auto loss prediction tasks.

In head to head trials, with only two non-PII inputs, the prediction has proven to be equal to or better than in production models.

Priced on a per API call basis with no access fees.

Fine-tuned Model

The model is first fine-tuned on a training dataset before providing a prediction based on the inputs provided.

Proven use cases for loss prediction, conversion, loss ratio improvement, and fraud detection.