Healthcare Insurance Verification & RCM Automation
Automated 300 patients/day, replaced 7 FTEs
Industry
Healthcare / RCM
Service
AI Integration & Agentic Workflows
Client
Healthcare Billing & RCM Company (US/India)
Timeline
4 weeks
The Problem
In the US healthcare system, clinics and doctors claim insurance, which requires a heavy, manual verification setup.
The client, an Indian company handling RCM (Revenue Cycle Management) and billing for US patients, had a large manual team solely dedicated to verifying insurance data.
Verification requires logging into various disparate payer portals to check if a patient's insurance is active, and extracting complex data like copay, coinsurance, and deductibles, making it a highly labor-intensive and error-prone process.
The Approach
EHR Integration
Built the foundational integration with eClinicalWorks (eCW) to automatically extract daily appointment schedules and patient data for specific dates, eliminating the need for manual reports.
Multi-modal Extraction Engine
Developed a dynamic extraction pipeline that adapts to the specific payer portal. Given the variance in portal designs, the system intelligently uses a combination of direct APIs, HTML parsing, Selenium-based web automation, and Chrome extensions to access the data.
AI Agentic Parsing
Implemented an AI agent layer using LLMs with JSON structured outputs to parse the highly irregular portal data (whether from HTML DOMs or API payloads) and reliably extract the exact copay, deductible, coinsurance, and active status.
Closed-loop Automation
Orchestrated the final step of the pipeline to take the structured verification results and automatically write them back into the patient records within eClinicalWorks via UiPath and APIs.
The Results
Reassigned to new clients (>50% of team)
Patients verified automatically per day
From idea to production pipeline
Key Takeaways
- →Healthcare automation often requires a 'by-any-means-necessary' approach to integration. Relying purely on APIs will fail when portals lack them; you must fallback to DOM parsing or RPA orchestration gracefully.
- →Agentic AI transforms web scraping. Instead of maintaining brittle regex patterns for 50 different insurance portals, an LLM can understand the implicit structure of the portal data and extract the necessary JSON schema.
Related case studies
AI SEO Agent: Internal Linking & Backlink Graph
Built a dual-agent LangGraph system using Firecrawl and Python to autonomously map, graph, and optimize internal linking structures and backlink quality for a 400+ page website.
Autonomous Inbound Sales & Scheduling Agent
Engineered an autonomous email agent using Gmail AI/PubSub, Perplexity, and OpenAI to research inbound leads, check calendars, and draft personalized replies instantly, eliminating manual work for the C-suite.