Building a Personalized Telegram Chatbot with Python and Gemini 2.0 Flash

🚀 Introduction

Creating a chatbot that feels like a real friend is an exciting challenge! In this guide, I’ll walk you through how I built a Telegram chatbot using Python and Google’s Gemini 2.0 Flash API. My goal was to make it respond in my personal style, switching between English and Vietnamese, and sometimes giving mentorship on programming like a senior developer.

---

🔹 Step 1: Setting Up the Telegram Bot

1️⃣ Create a Telegram Bot with @BotFather

1. Open @BotFather on Telegram.

2. Type: /newbot.

3. Follow the instructions & give your bot a name.

4. Save the Bot Token (We’ll use it later in our code).

---

🔹 Step 2: Get the Gemini API Key

1. Go to Google AI Studio.

2. Sign in & create a new API key.

3. Save this API key for later use.

---

🔹 Step 3: Install Required Libraries

Run the following in your terminal:

pip install python-telegram-bot requests python-dotenv

---

🔹 Step 4: Store Configurations Securely

1️⃣ Create a .env File

Create a file named .env in your project directory and add:

TELEGRAM_BOT_TOKEN=your-telegram-bot-token
GEMINI_API_KEY=your-gemini-api-key

This keeps your credentials safe and prevents them from being hardcoded.

2️⃣ Create responses.json for Predefined Replies

To make the bot more responsive, we define common replies in a JSON file. Create a new file responses.json and add various responses based on user interactions.

---

🔹 Step 5: Writing the Python Bot

Develop a Python script to handle messages, fetch AI-generated replies, and respond naturally.

---

🔹 Step 6: Running Your Bot

1. Replace YOUR_TELEGRAM_BOT_TOKEN with your actual bot token.

2. Replace YOUR_GEMINI_API_KEY with your Gemini API key.

3. Save the file as bot.py.

4. Run it with:

    python bot.py
   

5. Open Telegram and chat with your bot! 🚀

---

🔹 Next Steps

• Deploy it on a server (e.g., AWS, Heroku, or PythonAnywhere).

• Improve personality by feeding more of your own replies.

• Add voice & image message handling.

• Train it to remember past conversations better.

This was a fun project, and the bot now feels more real, engaging, and useful! Want help with deployment? Let me know! 😊