Workers' Comp Insurer Improves Reserve Accuracy With AI Prediction Model
The MEMIC Group has improved the accuracy of workers' compensation claims reserves following a six-month deployment of Gradient AI's Total Incurred Prediction model. The system predicts total claim costs earlier in their lifecycle, giving adjusters data-driven guidance for reserve decisions.
MEMIC, which covers workers' compensation across all 50 states, implemented the model to sharpen reserve accuracy at the individual case level. The company's conventional reserving methods worked, but lacked precision where it mattered most-in setting reserves for specific claims.
Results and Operational Impact
The six-month analysis showed reserve adjustments informed by the TIP model led to more timely updates to Experience Modifications, or X Mods-calculations that determine workers' comp pricing strategies. Improved accuracy in experience-based pricing at renewal followed.
Matt Harmon, Senior Vice President of Claims for MEMIC, said the company went live with the enhanced process in May 2025. "Reserve adjustments driven by TIP insights resulted in more timely updates to X Mods and improved accuracy in experience-based policy pricing at renewal," he said.
Beyond reserves, MEMIC identified additional uses for the model across claims and underwriting. These include evaluating open claims at the policy level when accounts leave MEMIC, analyzing claims within a policy to inform renewal pricing, and assessing catastrophic loss events for earlier loss severity estimates.
The tool also reviews claims across multiple policies when evaluating potential commutations or settlements with reinsurers. Underwriting teams gained earlier visibility into favorable X Mod trends and stronger alignment between claims outcomes and underwriting results.
Revenue and Cultural Shift
The TIP model contributed meaningfully to MEMIC's revenue growth in the first quarter of 2026. Improved reserve accuracy gave underwriters greater confidence in renewal pricing decisions, supporting more appropriate premium levels that reflect underlying risk more accurately.
Beyond financial gains, embedding predictive AI Data Analysis into daily decision-making shifted MEMIC toward a more proactive, data-driven culture. The approach distinguishes the company from competitors.
"Improved reserve accuracy has given MEMIC underwriters greater confidence in renewal pricing decisions," Harmon said. "In many cases, this has supported more appropriate premium levels that more accurately reflect underlying risk."
Implementation Approach
MEMIC took a measured, cross-functional approach to the deployment. Underwriting, claims, and actuarial leadership collaborated to evaluate the solution's value. Stan Smith, CEO of Gradient AI, said the goal was to integrate the model into MEMIC's broader claims assessment strategies, not simply deliver a model.
"This thoughtful, collaborative approach has helped MEMIC steadily build value via the TIP model even while it maintains its day-to-day focus on execution," Smith said.
Learn more about AI for Insurance applications and how carriers are deploying machine learning in claims operations.
Your membership also unlocks:
