🍵 TeaBuddy: An AI Assistant for Matcha & Healthy Drinks

TeaBuddy is a lightweight, fully offline AI chatbot built with Streamlit and powered by Ollama + Gemma 1B. It's your personal tea expert. This will help you to explore matcha, compare green teas, and answer any question about healthy drinks.

Designed for fast local interaction, TeaBuddy runs entirely on your machine. No OpenAI keys. No cloud APIs. Just our tea talk.

---

Current Features

• Natural tea chat interface (matcha, green tea or any healthy drinks)
• Runs locally using Ollama (without internet)
• Smart, themed replies + random emoji reactions
• Chat history + export to .txt
• Auto-clearing input after each question
• Lightweight UI with clean layout (used Streamlit)
• Tuned system prompt to stay helpful but flexible in diffrent version (not just matcha!)
• Works on any machine with Python + Ollama (mine is tested with Mac)

---

Chatting Demo

Watch this 38s demo or check demo.gif in this repo.

---

Project Structure

TeaBuddy/
├── app.py                # app file for Streamlit
├── requirements.txt      # Python dependencies
├── README.md
└── data/
     └── tea_knowledge_base.md   # (reference contents for future versions)          

---

Run TeaBuddy Locally

Requires Python 3.10+ and Ollama installed with a local model (e.g. gemma:1b).

1. Clone the repo

git clone https://github.com/saifar-tug/TeaBuddy.git
cd TeaBuddy

2. Set up your environment

python3 -m venv venv
source venv/bin/activate  # Mac/Linux
# OR
venv\Scripts\activate     # Windows

pip install -r requirements.txt

3. Pull a model with Ollama

ollama pull gemma:1b

4. Run TeaBuddy

streamlit run app.py

Visit: http://localhost:8501

---

Save Chat History

Click "Save chat to file" to export your conversation as a timestamped .txt file.

---

Built With

• Streamlit — Python Web App Framework
• Ollama — Local LLM AI Assistant
• gemma:1b — Lightweight local model (good balance of speed + size)

---

My Note

As a tea lover and AI enthusiast, I wanted a minimal, local-first AI assistant that doesn’t need a cloud API key to chat. I like the way of learning and experimenting with local LLMs, Streamlit, and quick prototyping.

---

For Reviewers

• built for portfolio
• fully runnable locally (offline)
• see the demo and code
• reach out if you would like me to walk you through the project live

---

Future Ideas

• Integrate more models (like LLaMA 3 or Mistral)
• Use embeddings + vector search for tea docs
• Test mode (funn quizzes on tea knowledge)
• Responsive layout for mobile use

---

License

This is one of my solo projects, fell free to use or adapt.

---

Got a tea question for TeaBuddy or LLM idea? Feel free to collaborate via Issues or Pull Requests!