AI Memory Assistant

Overview

AI Memory Assistant is a personal memory-enabled chatbot built using LangChain, FAISS, Hugging Face Embeddings, and Groq LLMs.

Unlike traditional chatbots, this assistant can remember important information shared by users, retrieve relevant memories using semantic search, and generate personalized responses.

The project demonstrates concepts such as:

• AI Agents
• Long-Term Memory
• Vector Databases
• Embeddings
• Retrieval-Augmented Generation (RAG)
• Large Language Models (LLMs)

---

Architecture

User
 │
 ▼
Memory Agent
 │
 ├── Memory Retrieval
 │         │
 │         ▼
 │      FAISS Vector Store
 │
 ├── Memory Storage
 │         │
 │         ▼
 │      FAISS Vector Store
 │
 ▼
Prompt Builder
 │
 ▼
Groq LLM (Llama 3.1)
 │
 ▼
Response

---

Workflow

Step 1: User Input

The user enters a message.

Example:

My name is Rahul

Step 2: Memory Retrieval

The query is converted into embeddings and compared against stored memories using FAISS.

Step 3: Prompt Construction

Relevant memories are added to the prompt along with the user's current message.

Example:

Relevant Memories:
My name is Rahul

Current User Message:
What is my name?

Step 4: LLM Response Generation

The prompt is sent to the Groq-hosted Llama 3.1 model.

The model generates a contextual response.

Example:

Your name is Rahul.

Step 5: Memory Storage

Important user information such as:

• Name
• Interests
• Preferences
• Goals
• Skills

is automatically stored for future retrieval.

---

Technologies Used

• Python
• LangChain
• Groq API
• Llama 3.1 8B Instant
• FAISS Vector Database
• Hugging Face Embeddings
• Sentence Transformers
• dotenv

---

Project Structure

AI-Memory-Assistant/
│
├── app.py
├── agent.py
├── memory_store.py
├── tools.py
├── requirements.txt
├── .env.example
├── .gitignore
└── README.md

---

Installation

Clone the repository:

git clone https://github.com/your-username/AI-Memory-Assistant.git
cd AI-Memory-Assistant

Install dependencies:

pip install -r requirements.txt

---

Environment Variables

Create a .env file in the root directory:

GROQ_API_KEY=your_groq_api_key_here

Get a free API key from Groq Console.

---

Running the Project

Start the application:

python app.py

Example:

You: My name is Rahul

Assistant: Nice to meet you Rahul.

You: What is my name?

Assistant: Your name is Rahul.

---

Key Concepts Demonstrated

AI Agent

The Memory Agent coordinates memory retrieval, memory storage, prompt generation, and response generation.

Embeddings

Text is converted into vector representations using Hugging Face sentence-transformer models.

Vector Database

FAISS stores embeddings and performs similarity search.

Retrieval-Augmented Generation (RAG)

Relevant memories are retrieved from FAISS and injected into the prompt before sending it to the LLM.

Semantic Search

The assistant can find related memories even when exact keywords do not match.

---

Future Improvements

• Streamlit Web Interface
• Multi-user Memory Support
• Memory Categorization
• Memory Importance Scoring
• Conversation History Tracking
• Database Integration (MongoDB/PostgreSQL)

---

Author

Developed as a mini project to demonstrate AI Agents, Memory Systems, Vector Databases, and Retrieval-Augmented Generation using LangChain.