Chat Therapy Bot

This project implements a therapy-style conversational agent using Streamlit and Groq’s LLaMA-3 models. It provides structured, safe, and predictable responses appropriate for mental-wellbeing use cases.

---

Design Flow (Figma)

Figma Board

https://www.figma.com/board/r2qgsg2mWeQp1G30gUHVQl/Chat-Therapy-Bot-Process-Flow

Features

Therapy-Style Chat

Focused, supportive conversation flow delivering short, empathetic responses designed for general mental wellbeing.

Crisis Detection

Monitors user messages for crisis keywords and activates an in-app crisis banner with guidance to immediate real-world help resources.

Scope Control

Strictly prevents answers outside mental-wellbeing scope. Topics related to medical advice, legal, finance, travel, academic work, coding, or operational instructions are refused with a standard supportive message.

PDF and Markdown Export

Allows users to export the entire chat session either as a formatted PDF or clean Markdown transcript.

Local Session Memory

Session content is stored only in Streamlit session memory. No persistent storage or logging.

---

Project Structure

genAI_ChatTherapy_bot/
│
├── app2.py               # Main Streamlit UI
├── agent.py              # Groq model wrapper and system prompt
├── requirements.txt      # Python dependencies
├── .env                  # GROQ_API_KEY (ignored in git)
│
└── assets/               # Icons + PDF font
    ├── green.png
    ├── red.png
    ├── favicon.png
    └── DejaVuSans.ttf

---

Installation

1. Create a virtual environment

python -m venv venv
source venv/bin/activate   # Linux/Mac
venv\Scripts\activate      # Windows

2. Install dependencies

pip install -r requirements.txt

3. Add the Groq API key

Create a .env file:

GROQ_API_KEY=your_key_here

---

Running the Application

streamlit run app2.py

Then open:

http://localhost:8501

---

How It Works

app2.py

• Handles full UI layout
• Implements crisis detection logic
• Renders chat messages with avatars
• Builds the model prompt
• Triggers PDF and Markdown exports
• Stores message history in session_state

agent.py

• Wraps Groq model client
• Contains the complete system prompt
• Enforces safety and refusal logic
• Streams the model’s token output

---

Safety Rules

The model is constrained to mental-wellbeing support only. It refuses to engage in:

• Medical diagnosis, treatments, prescriptions
• Legal, financial, travel, or technical advice
• Violence-related or harmful content
• Operational or actionable instructions

Crisis-related content triggers an emergency-help banner.

---

Exporting Chats

PDF

Uses FPDF with a clean page layout and a Unicode-compatible font.

Markdown

Simple, readable transcript for journaling or personal record keeping.

---

Author

Venkata Pavan Kumar Alapati MS Data Analytics, Clark University Creator of AIIDA, Stomes, and multiple agent-based AI systems