Palm Mind Interview Task 2

RAG-based agentic system -- LangChain + Redis + SMTP + PineCone

🗨️ FastAPI Chat & Appointment Booking System

A FastAPI-based chatbot with Redis-powered conversation history and an integrated appointment booking flow.
Supports name, email, phone, date, time, and message collection with validation for:

• ✅ Email format
• ✅ Phone format
• ✅ Date parsing (must be in the future)
• ✅ Time parsing (must be between 9 AM and 5 PM, and later than current time if booking is today)
• ✅ Automatic confirmation emails to customer and admin
• ✅ Updates booking info to pinecone database with embedding (sentence-transformer)

---

🚀 Features

• Natural conversation with memory stored in Redis
• Trigger words detection for booking (e.g., "book appointment", "schedule meeting", "reserve", "appointment", etc.)
• Step-by-step booking form through chat
• Smart date & time validation using dateparser and Python’s datetime
• Email notifications using SMTP
• Session handling for multiple users
• Clear session on "bye"

---

📦 Installation

1️⃣ Clone the repository

git clone https://github.com/pratut/interview_task_2.git
cd interview_task

2️⃣ Create and activate a virtual environment

python -m venv .venv
source .venv/bin/activate 

3️⃣ Copy environment template and configure

cp .env_template .env

4️⃣ Install dependencies

pip install -r requirements.txt

▶️ Run the application

uvicorn main:app --reload

API Endpoints

POST /chat

Request

{
    "session_id": "user123",
    "question": "Book appointment"
}

Response:

{
    "answer": "Sure! What's your name?"
}