The situation
An AI workflow for a medical-research platform: submitted manuscripts are read by Claude, structured into project fields, and filed into a consistent Google Drive hierarchy, with error branches when extraction fails.
All case studies
Automation
AI Manuscript Intake Automation
Claude reads research manuscripts, extracts metadata, and files them into structured Drive folders.
n8nClaude / AnthropicGoogle SheetsGoogle Drive
The problem
Research teams were opening every manuscript by hand, re-typing authors and study details, and filing inconsistently across Drive.
What we built
n8n reads pending rows from a master hub sheet, converts DOCX to text, calls Claude for metadata extraction, writes structured fields back to the sheet, and creates a seven-folder Drive tree per project.
What the system does
- Claude API metadata extraction (authors, study type, publication details)
- Structured write-back to the InsightMed master hub
- Automatic Google Drive folder hierarchy (protocol → archive)
- Error branches for failed extraction or parse
How it’s wired
The production n8n workflow , triggers, retries, and write-backs , not a slide-deck diagram.
Similar systems
All case studiesRelaydeskProduct / Internal Build
Relaydesk: WhatsApp Automation SaaS
Multi-tenant messaging architecture with queues, webhooks, and campaigns.
Next.jsTypeScriptTailwind
Read case study CITA MonitorWorking Software
CITA Monitor: Appointment Availability & Auto-Booking
Working software that watches ICP+ slots and alerts over Telegram.
PythonFastAPISQLite
Read case study WhatsApp AI Customer Response AutomationAutomation
WhatsApp AI Customer Response Automation
Inbound WhatsApp webhooks with AI first-touch replies and escalation.
WhatsApp APIWebhooksn8n
Read case study