Quotation Processor

End-to-end quotation automation for a trading company. An n8n workflow monitors IMAP for incoming RFQ emails, triggers a FastAPI backend to extract line items from PDF attachments using PDFPlumber, matches product codes against a catalogue via semantic similarity search over 1,000+ SKUs, generates formatted quotation PDFs with ReportLab, and delivers outputs to email and WhatsApp under 2 seconds.

Previously a sales assistant manually read each RFQ, looked up prices, typed a quotation, and sent it. Roughly 2 hours per batch which is now 30 seconds, fully unattended.

The vector-based SKU mapping engine handles messy customer part numbers (abbreviations, typos, alternative codes) that would break exact-match lookup. Containerised with Docker for deployment on a company VPS.