US not-for-profit health insurer saves $1.4M and cuts triage team 75% with gen AI automation

A large not-for-profit health insurer in New York, New Jersey, and Connecticut faced overwhelming case volumes for appeals and grievances triage. Data originated from multiple disparate channels and systems, requiring over 20 FTEs to manually categorize cases. Manual interpretation of medical records and regulations led to inconsistencies, errors, backlogs, and risk of missing critical turnaround times.

Cognizant developed a gen AI-powered Appeals & Grievances categorization assistant leveraging intent and entity recognition to extract information from structured and unstructured documents. The system uses dynamic knowledge mapping to align cases with relevant regulations and policies, and provides decision support by predicting category, subcategory, priority, and performing duplicate checks — automating the end-to-end triage workflow.

The solution delivered $1.4 million in cost savings over three years through rapid FTE reduction from 20 to 5 in just seven months. The automated triage process achieved a 90% accuracy rate in case categorization, significantly reducing backlogs and improving member response times.

• Generative AI can automate high-volume, judgment-intensive administrative workflows — reducing large manual teams by 75% within months.
• Combining intent recognition with dynamic regulatory knowledge mapping is key to achieving high accuracy in compliance-sensitive healthcare triage.
• Framing automation around decision support (rather than full replacement) enables rapid adoption and measurable cost savings.