Build an AI Memory System for Chatbots with Gemini in 30 Minutes

Overview

Ever frustrated when your AI chatbot forgets everything after a conversation ends? Today you'll build a memory system for Gemini that persists key details between chats - perfect for customer support bots, personal assistants, or creative collaborators. We'll use Google's Gemini API to create a simple but powerful context-preserving layer.

What You Will Be Able To Do

Why This Matters

Imagine you're a freelance designer using Gemini for client consultations. Today, you spend 10 minutes re-explaining the client's brand colors and preferences every new chat. With this system, Gemini will remember "Client X prefers minimalist designs in navy blue" automatically.

For customer support teams, this means no more asking customers to repeat order numbers. For personal use, your AI journaling companion remembers your important life events. Context persistence is what separates toy chatbots from truly useful AI agents.

Getting Started

pip install google-generativeai python-dotenv

Step-by-Step Guide

Step 1: Get Your Gemini API Key

1. Go to https://aistudio.google.com/app/apikey
2. Click "Create API key"
3. Copy the key and store it securely
4. Create a .env file in your project folder with:

GEMINI_API_KEY=your_key_here

Step 2: Set Up the Memory Database

Create a new Python file called memory_bot.py with:

python
import json
import os
from pathlib import Path
MEMORY_FILE = Path("ai_memory.json")
def load_memory():
    if not MEMORY_FILE.exists():
        return {}
    with open(MEMORY_FILE, 'r') as f:
        return json.load(f)
def save_memory(memory):
    with open(MEMORY_FILE, 'w') as f:
        json.dump(memory, f, indent=2)

Step 3: Connect to Gemini with Memory

Add this to your file:

python
import google.generativeai as genai
from dotenv import load_dotenv
load_dotenv()
genai.configure(api_key=os.getenv('GEMINI_API_KEY'))
model = genai.GenerativeModel('gemini-pro')
def chat_with_memory(user_id, message):
    memory = load_memory()
    user_memory = memory.get(user_id, {"history": []})
    
    # Build context from memory
    context = ""
    if user_memory["history"]:
        context = "Previous conversation:\n" + "\n".join(
            f"{msg['role']}: {msg['content']}" 
            for msg in user_memory["history"][-3:]
        ) + "\n\n"
    
    response = model.generate_content(
        context + f"User: {message}\nAI:"
    )
    
    # Update memory
    user_memory["history"].extend([
        {"role": "user", "content": message},
        {"role": "AI", "content": response.text}
    ])
    memory[user_id] = user_memory
    save_memory(memory)
    
    return response.text

Step 4: Test Your Memory Bot

Add this test code:

python
if __name__ == "__main__":
    print("Memory Bot activated! Type 'quit' to exit.")
    user_id = input("Enter your user ID: ").strip()
    
    while True:
        message = input("You: ").strip()
        if message.lower() == 'quit':
            break
        print("AI:", chat_with_memory(user_id, message))

Step 5: Run and Experience Persistent Memory

1. Run python memory_bot.py
2. Enter a test user ID
3. Have a conversation
4. Quit and restart - your history will remain!

Real-World Example

Sarah, a nutrition coach, uses this to track client progress. When Client123 asks "What was my calorie target?", Gemini recalls from their last session: "Based on our 3/15 chat, your target is 1800kcal with 30% protein." Sarah saves 15 minutes per client by eliminating repetitive explanations while maintaining personalized service.

Common Mistakes to Avoid

Pro Tips

Your Challenge

Summary

You've built what most AI tools lack - persistent memory. This foundation lets you create truly personalized AI experiences that evolve over time. Next, try connecting this to a Discord bot or adding automatic memory categorization.