TIPS ON HOW TO MAKE A CHATBOT USING FREE GEMINI API KEYS

Source: Dev.to

Overview

The Gemini API provides cost‑effective access to Google’s most capable AI models. This guide outlines the architecture and implementation steps for building a chatbot while keeping your API keys secure.

Architecture

Setup

1. Obtain an API key from the Google AI Studio.

2. Create a Python environment and install the required library:

   pip install -q -U google-generativeai

3. Store the API key in a .env file and add the file to .gitignore to prevent accidental exposure.

Initializing the Model

# service.py or views.py
import google.generativeai as genai
import os

# Securely load your API key
genai.configure(api_key=os.environ["GEMINI_API_KEY"])

# Initialize the model (Gemini 2.5 Flash is recommended for speed)
model = genai.GenerativeModel('gemini-2.5-flash')

Creating a Chat Session

The standard prompt‑response flow is stateless. To maintain conversation history, use the .start_chat() method.

# Start a chat session with an empty history
chat = model.start_chat(history=[])

def get_chatbot_response(user_input: str) -> str:
    """Send a message to the model and return the text response."""
    response = chat.send_message(user_input, stream=False)
    return response.text

Safety Settings

Gemini includes built‑in filters for harassment and dangerous content. You can adjust these settings in the configuration if the model is overly restrictive for your use case.

System Instructions (Persona)

Define a system instruction to give the chatbot a consistent persona:

system_prompt = (
    "You are a professional first‑aid assistant. "
    "Provide clear, step‑by‑step emergency instructions."
)

# Example of initializing with a system instruction
model = genai.GenerativeModel(
    'gemini-2.5-flash',
    system_instruction=system_prompt
)

Error Handling

Wrap API calls in try‑except blocks to handle rate limits, quota errors, and network timeouts.

def safe_get_response(user_input: str) -> str:
    try:
        return get_chatbot_response(user_input)
    except Exception as e:
        # Log the error and return a friendly message
        print(f"Error: {e}")
        return "Sorry, I'm experiencing technical difficulties. Please try again later."

Security Best Practices

• Never call the Gemini API directly from the frontend (JavaScript). Exposing the API key in the browser allows anyone to steal it and consume your quota.
• Always route requests through your backend, where the API key is stored securely in environment variables.
• Keep the .env file out of version control (.gitignore).

By following these guidelines, you can build a functional and secure chatbot using free Gemini API keys.