AI Workflow & Bus Management System

Python FastAPI JavaScript LangChain Google Sheets API SMTP

View Live Demo

The Challenge

Managing bus logistics manually was slow and error-prone. Coordinating approvals, drafting professional emails, and maintaining accurate records across spreadsheets consumed valuable time and introduced inconsistencies.

I built an intelligent system that automates 90% of the workflow—from processing requests to generating context-aware responses. The AI chatbot understands policy documents and queries the system's spreadsheet in real-time, providing instant answers to logistics questions.

Key Features

AI Assistant for Policy Queries: The chatbot works directly with a Google Spreadsheet, querying real-time data using Python and LangChain to answer questions about routes, schedules, and policies.
Automated Email Draft Generation: The system generates professional, context-aware email drafts based on request details, saving hours of manual writing and ensuring consistency.
Two-way Sync with Google Sheets: Seamless integration allows the system to read requests from Google Forms, process them through the AI pipeline, and update approvals back to the spreadsheet automatically.
Custom Dashboard for Approvals: Built with JavaScript and FastAPI, the dashboard provides approvers with a clean interface to review AI-generated content and make decisions efficiently.
Real-time Monitoring: A QA bot continuously monitors the sheet for changes, updating the system state and triggering notifications based on approval status.

Technical Implementation

The system architecture combines several modern technologies: FastAPI serves as the backend API, handling authentication and business logic. LangChain orchestrates the AI pipeline, connecting the chatbot to the Google Sheets data source for retrieval-augmented generation.

The frontend dashboard uses JavaScript for a responsive user experience, while Python scripts handle the heavy lifting of spreadsheet synchronization and email automation via SMTP. The entire system is deployed on Render, ensuring reliable uptime and scalability.