Archived:

A structured AI memory system powered by Retrieval-Augmented Generation (RAG) and graph-based memory storage.

---

Overview

Archived is an AI-driven memory architecture that transforms user interactions into a persistent, structured knowledge graph. It allows an LLM (Gemini API) to store, retrieve, and reason over past information like a long-term memory system.

Instead of treating each query independently, Archived builds continuity across conversations.

---

Core Concept

Archived turns unstructured input into structured memory:

User Input → Ingestion → Memory Graph → Retrieval → Reasoning (Gemini) → Response

Each interaction becomes a node, connected through semantic relationships, enabling long-term contextual awareness.

##System Architecture

1. Ingestion Agent

• Extracts entities, keywords, and metadata
• Converts raw input into structured memory nodes

2. Organizer Agent

• Maintains graph structure
• Creates relationships between nodes

3. Retrieval Agent

• Searches memory graph for relevant context
• Returns top-k relevant nodes

4. Reasoning Agent

• Uses Google Gemini API as the reasoning engine
• Combines retrieved memory + new input
• Generates structured response

5. Evolution Agent

• Node importance is updated based on how frequently the user interacts with it.
• Frequently accessed nodes gradually strengthen over time. *High-importance nodes may be promoted to long-term memory. *Infrequently accessed nodes slowly fade in relevance.

---

Why This Matters

Archived demonstrates how AI systems can evolve from stateless chatbots into memory-aware agents capable of:

• Personalized reasoning
• Context retention over time
• Structured knowledge accumulation

---

🔮 Future Improvements

• Vector database integration
• Semantic embeddings for retrieval
• Memory decay / importance scoring
• Visual graph UI for memory exploration
• Multi-user memory isolation

---

🧠 Philosophy

“A model is only as intelligent as what it remembers.”

Archived focuses on giving AI systems a structured long-term memory layer, bridging reasoning and persistence.

---

⚙️ Installation & Setup

1. Clone the Repository

git clone https://github.com/pranalisree/archived.git
cd archived

2. Create Virtual Environment (Recommended)

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

3. Install Dependencies

pip install -r requirements.txt

If requirements.txt is not available, install core dependencies manually:

pip install google-generativeai numpy networkx

---

4. Set Up Gemini API Key

You must configure your Google Gemini API key:

Create a .env file in the root directory:

GEMINI_API_KEY=your_api_key_here

Or export it dir

MIT License