Using Jev to Reduce Our Reasoner Cost by 30%
A look at using Jev for document routing, source recall, and cost reduction inside a production insurance claims AI workflow.
At Altis, a claims agent for property and casualty providers is being developed to analyze policy documents, OEM repair procedures, and insurer guidelines. The current process reads all documents before making recommendations, resulting in high costs and low efficiency. By introducing Jev, a specialized model, the insurance company aims to reduce the cost of their reasoner by 30%.
Jev is a System One model designed for making typed probabilistic decisions, trained using Reinforcement Learning for Calibrated Decisions (RLCD). The model takes a technical source and outputs a decision based on whether it contains relevant information for the repair case. By using Jev, the insurance company can reduce the number of documents analyzed while still retaining the necessary information, resulting in significant cost savings.
Written by urgent.news from HackerNoon's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.