How Max Barinov Cut AI Token Consumption by 10x Analyzing Medical Records at Adentris

This article profiles how Adentris tackled the challenge of processing massive electronic medical records using LLMs. Instead of a naive “process everything” approach, the team built a multi-agent system that breaks EMRs into manageable chunks and tracks only incremental changes. The result was a 10x reduction in token usage, faster processing, and successful HIPAA compliance audits—showing that architecture, not model size, is the key to scalable healthcare AI.