🚀 I Built an AI-Powered Fest Assistant with Agents, RAG & Planning (Pragyan @ NITT)

I deleted Instagram more than a year ago, and honestly, it saved me from a lot of distractions.

But there was an unexpected downside.

A lot of informal, real-time information — especially during college events — still lives there.

During our college fest, for example:

• Event schedules
• Last-minute updates
• Food stall announcements
• Informal activities

…everything gets posted on Instagram.

At the same time:

• Fest details and significance are on the official website
• Food stall info is on a separate app
• The entire 3-day schedule is compressed into a few posts

There’s no single place to get a clear, structured view of everything.

And that’s when it hit me:

Most college fests have websites.
Some even have apps.

But none of them actually help you navigate the fest intelligently.

They give information.
They don’t give guidance.

But I wanted to build something smarter —
an AI assistant that actually understands queries, plans your day, and even helps you find teammates.

So, I built Pragyan Mentor Assistant — an AI-powered system for navigating a techno-managerial fest.

🎯 Problem

During college fests like Pragyan (NIT Trichy):
There are

• There are dozens of events, workshops, and shows
• Information is scattered across PDFs, sites, and posters

• Users don’t know:

  • what to attend
  • what matches their interests
  • how to plan their time
  • who to team up with

👉 Traditional apps = static information
👉 I wanted intelligent interaction

💡 Solution

I built a multi-tool AI assistant that can:

• 🎉 Answer questions about events, workshops, proshows
• 🍔 Show food stalls & mess menu
• 🧠 Recommend activities based on user intent
• 📅 Plan your schedule
• 🤝 Match you with like-minded participants/Suggest potential teammates (prototype)
• 📚 Answer fest-related questions using RAG

🧠 System Design

Instead of a simple chatbot, I designed it as a tool-using agent system.

🔹 Tools

• fetch_events
• fetch_workshops
• fetch_food_stall
• fetch_mess_menu
• pragyan_bot (RAG-based)
• smart_recommender
• planner
• buddy_matcher

🔹 Agent Flow

1. User query
2. LLM decides:

• Which tool to call
  1. Tool executes
  2. Response is generated in natural language

📚 Retrieval Approach

This system uses a hybrid retrieval strategy at the system level:

• Structured retrieval (keyword-based)

  • Direct tool calls for events/workshops
  • Fast and deterministic

• Semantic retrieval (RAG)

  • Vector search over fest documents
  • Handles open-ended queries

👉 This combination allows both precision and flexibility

📚 RAG (Retrieval Augmented Generation)

To handle fest knowledge:

• Used:

  • Text files (events, shows, lectures, FAQs)

• Built:

  • FAISS vector store

• Retrieval:

  • Semantic search on query

• Response:

  • Context-aware answers

🧠 Memory

Using:

• InMemorySaver() (LangGraph)

👉 Enables:

• remembering user preferences
• better recommendations
• conversational continuity

🤖 Smart Features

🎯 Recommendations

Understands intent like:

"What should I attend if I like tech and fun?"

📅 Planner Agent

"Plan my next 3 hours"

Generates a structured schedule based on:

• time
• interests
• available events

🤝 Buddy Matching (Prototype)

Matches based on:

• interests
• level
• context (e.g. case study competitions)

Uses a small dataset to demonstrate logic

🖥️ UI

Built with Streamlit:

• Chat-based interface
• Quick action buttons
• Structured responses

🚀 Deployment

Deployed on Render (free tier)
Environment variables for API security

🎥 Demo

👉 https://www.loom.com/share/13f87025a9154a55b80fc240bfc91ba2

🛠️ Tech Stack

• Python
• LangChain
• OpenAI API
• FAISS
• Streamlit
• Render

⚠️ Challenges Faced

• RAG retrieval quality (chunking + parsing issues)
• Tool selection accuracy
• Structuring multi-agent workflow
• Deployment + API key handling

🔄 Ongoing Improvements

Some features I’m actively working on:

1. Adding database-backed user profiles for real buddy matching
2. Improving RAG with better retrieval and evaluation
3. Expanding dataset coverage for more complete fest information
4. Exploring true hybrid retrieval + reranking

📈 What I Learned

• Building agents > building chatbots
• RAG needs data structuring, not just embeddings
• UI matters a lot for perceived intelligence
• Deployment and debugging are part of the real challenge

🔗 Links

• 🔒 Live demo available on request

💭 Final Thoughts

This project made me realize:

👉 The future isn’t just about LLMs
👉 It’s about systems built around them

If you have suggestions or ideas to improve this, I’d love to hear them!