Automation
Automated Feedback Intelligence System
Built an end-to-end patient feedback processing system using n8n and Claude AI that classifies, prioritizes, and escalates complaints automatically. Includes a scheduled executive reporting workflow that aggregates weekly trends and delivers AI-generated operational summaries to clinic leadership.
Business Problem:
Healthcare clinics collect patient feedback at scale but lack a reliable system to process it consistently. Individual complaints arrive without context, patterns go undetected for weeks, and high-risk issues like medication errors or clinical failures sit in an unread spreadsheet until they become a larger problem. The operational cost is real: delayed responses, repeat complaints, and leadership making decisions without accurate data.
How it works:
The solution is a two-workflow automation built in n8n, powered by Claude 3.5 Haiku as the core analysis agent. Each submission triggers a deduplication check before the agent classifies sentiment, assigns a complaint category, and scores priority in a single structured pass. High-risk records route immediately to an escalation email; all records append to a results sheet that feeds a live Looker Studio dashboard.
A second workflow runs on a Monday morning schedule, reads the previous seven days of processed data, calculates aggregate metrics via a code node, and passes the summary to Claude 3.5 Sonnet to generate an executive report, which is delivered directly to the clinic manager. The full pipeline handles intake, classification, escalation, storage, and reporting with zero manual intervention between form submission and the leadership inbox.
Gallery
