Order Email Extraction
Reconstructs order and shipment state for a logistics workflow: reads a business mailbox, classifies each message by intent and extracts structured records without manual entry.
- Area
- AI · Operations & tools
- Main stack
- Angular · OpenAI API · LLM · Structured output
Key features
- • Mailbox retrieval through the Google APIs
- • Message classification into nine intent categories
- • Structured extraction constrained by a JSON schema
- • Six relational tables with an order event history
- • Web interface with order flow and customer list
About the project
Service that turns the order and shipment emails of a business into database records, so nobody retypes them. A Node.js backend reads the mailbox through the Google APIs and submits each message to a language model, which returns JSON constrained by a declared schema: intent, confidence and blocks for order, confirmation and shipment. Nine intent categories range from order requests and supplier confirmations to automatic replies. Results are written to six PostgreSQL tables, and an Angular front end shows the order flow and the customer list.
Technologies used
Angular 
OpenAI API 
LLM 
Structured output 
Node.js 
Express 
PostgreSQL 
Google APIs 
JSON Schema