Cutting Patient Intake Time by 68%

The Problem

Meridian Health operated 12 clinics with a paper-heavy intake process. Staff spent an average of 45 minutes per patient on data entry, insurance verification, and record matching. Error rates hovered at 11%, leading to billing disputes and delayed care.

Our Approach

We ran a two-week discovery sprint across three pilot clinics, mapping every intake touchpoint. We identified 23 discrete steps that could be augmented or fully automated with LLM-based extraction and classification. We allocated AI credits across document parsing, entity extraction, and verification modules.

The Solution

Built a multi-model pipeline: OCR for document scanning, an LLM for entity extraction and record matching, and a classification model for insurance routing. The system integrates with their existing EHR via HL7 FHIR APIs. A staff-facing dashboard shows real-time credit usage and processing status.

Results

• 68% reduction in intake processing time
• Error rate dropped from 11% to 2.3%
• Staff reallocated 340+ hours/month to patient care
• ROI positive within 8 weeks of deployment