Inspiration
So i noticed that when i went to the admin department that there was a lot of problem in getting the document approved as it goes through many steps and there is no transparency of where the document might have gotten stuck.
What it does
This provides an integrated system that links different departments, by first taking in an input user query, breaks it down to the required steps and which all departments it should go through. In the end it provides an estimate of how long it will take for the query to be resolved. There is also transparency as all the departments can see at which step is the document and where the bottle neck is coming. Along with that there is a feature for the admin to ask the user any extra details that they may require, making the entire process online and reduce travel time.
How we built it
Backend: FastAPI with RAG-based context retrieval for Gemini routing decisions AI integration: Gemini generates multi-department workflows with time estimates from RAG context Frontend: React app with real-time request tracking, timeline, and clarification center.
Challenges we ran into
RAG context retrieval — selecting relevant knowledge for accurate routing Multi-department routing complexity — ensuring valid, sequential workflows Real-time tracking — maintaining accurate state as requests move through departments
Accomplishments that we're proud of
AI-powered automatic routing that determines department workflows
What we learned
RAG implementation for context-aware AI decisions Gemini API integration with structured JSON responses
What's next for RequestFlow
Brining rate limiting and controls against abuse. Database integration (PostgreSQL/MongoDB) for persistence Expand department catalog and request types Authentication and role-based access control Analytics dashboard with visualizations Integration with existing DTU systems
Log in or sign up for Devpost to join the conversation.