AgentOS

Production-Grade AI Operating System

An extensible AI platform featuring agent orchestration, persistent memory, Retrieval-Augmented Generation (RAG), tool execution, semantic search, and streaming conversations.

Overview

AgentOS is a production-oriented AI Operating System designed to run intelligent AI agents with persistent memory, semantic retrieval, tool execution, streaming responses, and modular orchestration.

Unlike simple chatbot applications, AgentOS provides the backend infrastructure required to build real AI products.

It combines:

AI Agent Runtime
Long-term Memory
Semantic Search (RAG)
Tool Calling
Knowledge Base
Conversation Management
Multi-Provider LLM Support
Production Architecture

The project follows scalable software engineering principles inspired by modern AI systems used in production.

Demo Link:

Coming Soon

Live Demo

Frontend: https://agent-os-web-puce.vercel.app

Backend API: https://agentos-api-sx1h.onrender.com

GitHub: https://github.com/Ajayreddy18/AgentOS

Screenshots

Login	Dashboard
Organizations
Projects
Environments
Agents	Conversations
Knowledge	Prompts
Tools
Settings
Chat

Metrics

Audit

Why AgentOS?

Modern AI applications require much more than calling an LLM API.

Production AI systems need:

Agent orchestration
Persistent memory
Context retrieval
Tool execution
Multi-provider support
Conversation history
Knowledge management
Streaming responses
Modular architecture

AgentOS brings these capabilities together into one extensible platform.

System Architecture

                User
                  │
                  ▼
             React Frontend
                  │
                  ▼
          Express REST API
                  │
      ┌───────────┼────────────┐
      ▼           ▼            ▼
 Authentication Runtime   Conversations
      │           │            │
      ▼           ▼            ▼
 Memory     Retrieval      Tool Runtime
      │           │            │
      └──────► PostgreSQL ◄────┘
                 pgvector
                  │
                  ▼
            Groq / Jina APIs

Key Features

AI Runtime

AI Agent Runtime
Runtime Loader
Agent Configuration
Model Management
Provider Management
Environment Management

Authentication & Organization

JWT Authentication
User Management
Organizations
Projects
Environments

Conversation System

Chat API
Conversation History
Streaming Responses (SSE)
Context Window Management

Memory System

Long-term Memory
Conversation Memory
Memory Retrieval
Memory Manager
Context Injection

Retrieval-Augmented Generation (RAG)

Knowledge Base
Document Management
Chunk Storage
Embedding Generation
Semantic Retrieval
Vector Search using pgvector

Tool Calling

Tool Registry
Tool Runtime
Dynamic Tool Loading
Tool Executor
Built-in Tools

Example tools:

Calculator
Date & Time

Designed for easy extension with custom tools.

AI Providers

Current support:

Groq
Jina Embeddings

Provider architecture allows easy addition of:

OpenAI
Anthropic
Google Gemini
Ollama
Azure OpenAI

AI Planning

Planner
Tool Selection
Orchestration Pipeline
Runtime Context Assembly

Backend Architecture

Layered Architecture
Service Layer
Repository Pattern
Dependency Injection Style
Modular Components
Scalable Folder Structure

Technology Stack

Repository Structure: Monorepo

apps/ packages/ docs/

Backend

TypeScript
Node.js
Express.js
PostgreSQL
Drizzle ORM
pgvector
JWT Authentication
Zod Validation
Pino Logger

AI Stack

Groq API
Jina Embeddings
RAG
Vector Search
Tool Calling
Streaming Responses

Frontend

React
Vite
Tailwind CSS
TypeScript

DevOps

GitHub
Husky
ESLint
Prettier

Project Architecture

AgentOS

├── Authentication
├── Organizations
├── Projects
├── Environments
├── Providers
├── Models
├── Agents
├── Conversations
├── Runtime
├── Planner
├── Orchestrator
├── Memory
├── Retrieval
├── Knowledge Base
├── Documents
├── Embeddings
├── Tool Runtime
├── Streaming
└── API Layer

Implemented Modules

Module	Status
Authentication	✅
Users	✅
Organizations	✅
Projects	✅
Environments	✅
AI Providers	✅
Models	✅
Agents	✅
Runtime Loader	✅
Conversations	✅
Streaming Chat	✅
Memory Manager	✅
Retrieval	✅
Knowledge Base	✅
Documents	✅
Embeddings	✅
Vector Search	✅
Tool Runtime	✅
Tool Registry	✅
Tool Executor	✅
Planner	✅
Orchestrator	✅

API Highlights

REST APIs include:

Authentication
Users
Organizations
Projects
Environments
Providers
Models
Agents
Conversations
Chat
Streaming Chat
Documents
Knowledge Base
Embeddings

Repository Structure

AgentOS

apps/
    api/
    web/

docs/

packages/

Running Locally

Clone

git clone https://github.com/Ajayreddy18/AgentOS.git

cd AgentOS

Install

npm install

Configure Environment

Create:

apps/api/.env

Example:

DATABASE_URL=

JWT_SECRET=

GROQ_API_KEY=

JINA_API_KEY=

Run Backend

cd apps/api

npm run dev

Run Frontend

cd apps/web

npm run dev

Deployment

Frontend: Vercel

Backend: Render

Database: PostgreSQL

Vector Database: pgvector

LLM Provider: Groq

Embedding Provider: Jina AI

Status: Production Ready

Project Stats

Languages: TypeScript

Architecture: Monorepo

Modules: 20+

REST APIs: 50+

Database Tables: 20+

Production Features:

• Authentication • Organizations • AI Runtime • Persistent Memory • Knowledge Base • RAG • pgvector Search • Streaming Responses (SSE) • Tool Calling • Planner • Runtime Loader • Multi-provider LLM Support

Engineering Principles

This project emphasizes:

Clean Architecture
Modular Design
SOLID Principles
Separation of Concerns
Type Safety
Production Readiness
Scalability
Maintainability

Future Roadmap

Multi-Agent Collaboration
Background Agent Jobs
Agent Marketplace
Workflow Builder
Observability Dashboard
Runtime Inspector
Docker Deployment
Kubernetes Deployment
OAuth Authentication
Multi-LLM Routing
Human-in-the-Loop Approval
Plugin SDK
Monitoring & Metrics

Learning Outcomes

Building AgentOS involved implementing concepts including:

AI Agent Systems
Retrieval-Augmented Generation (RAG)
Vector Databases
Embedding Pipelines
Tool Calling
Streaming APIs
Semantic Search
REST API Design
Authentication
PostgreSQL
TypeScript Backend Architecture
Clean Software Engineering Practices

Author

Ajay Reddy

AI Engineer | Backend Engineer | Generative AI

GitHub: https://github.com/Ajayreddy18

LinkedIn: https://www.linkedin.com/in/ajayreddyofficial

Email: najayreddy2424@gmail.com

If you found this project interesting, consider giving it a ⭐.

Name		Name	Last commit message	Last commit date
.husky		.husky
apps		apps
assets/screenshots		assets/screenshots
docker		docker
docs		docs
packages		packages
.dockerignore		.dockerignore
.gitignore		.gitignore
.prettierignore		.prettierignore
README.md		README.md
docker-compose.app.yml		docker-compose.app.yml
docker-compose.yml		docker-compose.yml
eslint.config.mjs		eslint.config.mjs
package-lock.json		package-lock.json
package.json		package.json
prettier.config.mjs		prettier.config.mjs
project_structure.docx		project_structure.docx
tsconfig.base.json		tsconfig.base.json
tsconfig.json		tsconfig.json

Latest commit

History

Folders and files

Repository files navigation