Week Plan Chat

A LangChain + RAG-based web app to chat with your weekly goals.

Overview

Week Plan Chat is an AI-powered weekly planner that allows users to upload their weekly goal PDFs categorized as Work, Personal, and Health. Users can interact with them using natural language queries.
The app uses LangChain and RAG (Retrieval-Augmented Generation) to fetch relevant context from uploaded documents and answer questions conversationally.

Key Features:

• Upload 4 weekly schedule PDFs (Work Goals, Personal Goals, Health Goals).
• Chat with an LLM about your goals and schedules.
• Filtered Vector search with (based on week and question type) Qdrant for relevant context retrieval.
• Real-time answers via Server-Sent Events (SSE) streaming.
• Background jobs with Celery for heavy tasks.
• Cron jobs (Celery Beat) for scheduled tasks.
• OOP-based FastAPI architecture for cleaner, modular, and maintainable backend code.

---

Tech Stack

Frontend:

• React.js
• TailwindCSS
• Zod + React Hook Form (form validation)

Backend:

• FastAPI
• LangChain
• Qdrant Vector Database
• Supabase (auth + storage + database)
• Redis (caching + Celery broker)

Background & Scheduled Jobs:

• Celery (background processing)
• Celery Beat (cron jobs)

Other:

• Server-Sent Events (SSE) for streaming responses

---

How It Works

1. Upload PDFs – Users upload weekly goal PDFs for Work, Personal, and Health.
2. Process & Store – Documents are chunked, embedded, and stored in Qdrant for retrieval.
3. Ask Questions – Users type questions like:

   "What are my Monday work deadlines?"

4. Get AI Answers – The app retrieves relevant chunks and generates an LLM response in real time.

---

Installation

# Clone the repository
git clone https://github.com/yourusername/weekwise.git
cd week-plan-chat

# Backend setup
cd backend
pip install -r requirements.txt
uvicorn app.main:app --reload --reload-dir app

#running celery
celery -A app.config.celery_app.celery_app worker --beat --loglevel=info

# Frontend setup
cd frontend
npm install
npm run dev

---

Background & Cron Jobs

Background Jobs: Heavy document processing (chunking, embedding) is offloaded to Celery workers. Cron Jobs: Periodic tasks (e.g., sending weekly summaries or reminders) run on schedule via Celery Beat.

---

Windows Redis Setup (via Docker)

If you are running this project on Windows, it’s recommended to run Redis inside Docker instead of installing it manually.

Steps:

1. Make sure you have Docker Desktop installed and running.
2. Open a terminal (PowerShell or Command Prompt) and run:

# Pull the latest Redis image
docker pull redis

# Run Redis container
docker run --name redis-server -p 6379:6379 -d redis

# Stop Redis container
docker stop redis-server

# Start Redis container again
docker start redis-server

---

📈 Improvements

1. Put a limit on number of messages in a single chat.
2. Use langchain-qdrant package instead of qdrant package for better integration.