AIOS – Enterprise Multi-Agent RAG Assistant

AIOS is a scalable Multi-Agent Retrieval-Augmented Generation (RAG) platform designed to automate document retrieval, reasoning, and report generation. The system leverages a multi-agent architecture built using FastAPI, LangGraph, CrewAI, ChromaDB, and Gemini API to provide intelligent and context-aware responses.

---

Overview

AIOS combines multiple specialized AI agents that collaborate to process user queries efficiently. The platform retrieves relevant information from documents, performs reasoning, and generates structured responses using Large Language Models.

The project demonstrates concepts such as:

• Agentic AI
• Multi-Agent Systems
• Retrieval-Augmented Generation (RAG)
• Vector Databases
• LLM Integration
• API Development
• Scalable Backend Architecture

---

Key Features

• Multi-Agent Workflow Architecture
• Retrieval-Augmented Generation (RAG)
• FastAPI Backend
• ChromaDB Vector Database Integration
• Gemini API Powered Reasoning
• Modular Service-Oriented Design
• Context-Aware Information Retrieval
• Scalable and Extensible Architecture

---

Architecture

User Query
     │
     ▼
 FastAPI API
     │
     ▼
Supervisor Agent
     │
 ┌───┴─────────┐
 ▼             ▼
Retrieval   Report
 Agent       Agent
     │
     ▼
 ChromaDB
     │
     ▼
 Gemini API
     │
     ▼
 Final Response

---

Repository Structure

AIOS
│
├── app
│   ├── agents
│   │   ├── supervisor.py
│   │   ├── retrieval_agent.py
│   │   └── report_agent.py
│   │
│   ├── api
│   │   └── main.py
│   │
│   └── services
│       ├── vector_store.py
│       ├── gemini_service.py
│       └── pdf_loader.py
│
├── docs
│   └── architecture.md
│
├── .env.example
├── .gitignore
├── requirements.txt
└── README.md

---

Components

Supervisor Agent

Acts as the orchestrator of the system. It receives user requests and routes tasks to the appropriate agents.

Retrieval Agent

Responsible for retrieving relevant information from the vector database using semantic search techniques.

Report Agent

Processes retrieved information and generates structured reports and responses.

Gemini Service

Handles communication with Google's Gemini models for reasoning and content generation.

Vector Store

Manages document embeddings and semantic retrieval using ChromaDB.

PDF Loader

Loads and preprocesses PDF documents for indexing and retrieval.

---

Technology Stack

Programming Languages

• Python
• SQL

AI / GenAI

• LangChain
• LangGraph
• CrewAI
• Gemini API
• Retrieval-Augmented Generation (RAG)

Backend

• FastAPI
• REST APIs

Vector Database

• ChromaDB

Development Tools

• Git
• GitHub
• Docker

---

Workflow

1. User submits a query through FastAPI.
2. Supervisor Agent receives the request.
3. Retrieval Agent fetches relevant context from ChromaDB.
4. Report Agent processes the retrieved information.
5. Gemini API generates an intelligent response.
6. Final response is returned to the user.

---

Installation

Clone Repository

git clone https://github.com/pranaviredd/AIOS.git
cd AIOS

Install Dependencies

pip install -r requirements.txt

Configure Environment Variables

Create a .env file:

GEMINI_API_KEY=your_api_key

Run Application

uvicorn app.api.main:app --reload

---

Future Enhancements

• PDF Upload Interface
• Real-Time Document Indexing
• Multi-Document Retrieval
• Agent Memory Management
• MCP Integration
• Redis-Based Caching
• Docker Deployment
• Kubernetes Deployment
• Streamlit Dashboard

---

Learning Outcomes

This project demonstrates practical experience with:

• Multi-Agent AI Systems
• Agentic Workflows
• Retrieval-Augmented Generation
• Large Language Models
• Vector Databases
• Backend API Development
• Software Architecture Design

---

Author

Pranavi Reddy Miriyala

B.Tech – Artificial Intelligence and Data Science
VNR VJIET, Hyderabad

GitHub: https://github.com/pranaviredd

---

License

This project is licensed under the MIT License.