Agentic AI Workspace Platform featuring RAG, Memory, Research, Developer Tools, and Autonomous Workflows.
Omni AgentOS is a full-stack AI workspace designed to combine conversational AI, semantic memory, document research, developer tooling, and agentic workflows into a unified platform.
The system integrates Large Language Models (LLMs), Retrieval-Augmented Generation (RAG), vector memory, real-time streaming, web research, and workspace-aware developer tools to create an AI operating environment capable of assisting with coding, research, and knowledge management tasks.
- Real-time token streaming
- Context-aware conversations
- Multi-turn memory support
- Markdown and code rendering
- Repository exploration
- File reading and writing
- Workspace-aware AI assistance
- Terminal command execution
- ChromaDB vector storage
- Semantic memory retrieval
- Long-term conversational memory
- Memory search and management
- Web search integration
- Document upload and ingestion
- Contextual retrieval
- Research-focused AI workflows
- LangGraph orchestration
- Tool calling
- Workspace awareness
- Human-in-the-loop architecture
| Component | Purpose |
|---|---|
| Next.js Frontend | User interface and workspace |
| FastAPI Backend | API and orchestration layer |
| LangGraph | Agent execution framework |
| Groq LLM | Language model inference |
| ChromaDB | Vector memory storage |
| Sentence Transformers | Embedding generation |
| Tavily | Web research capabilities |
| WebSockets | Real-time communication |
- Next.js 15
- React
- TypeScript
- Tailwind CSS
- Zustand
- Framer Motion
- FastAPI
- LangGraph
- WebSockets
- Pydantic
- Groq
- ChromaDB
- Sentence Transformers
- RAG Pipelines
- Semantic Search
- Docker (Planned)
- CI/CD (Planned)
- GitHub
backend/
frontend/
docs/
├── architecture/
├── diagrams/
└── screenshots/
- AI Chat & Streaming Responses
- Developer Mode & Workspace Awareness
- Tool Calling (LangGraph integration)
- Memory System & Semantic Search
- Research Mode & Document Ingestion
- Settings & BYO API Key Multi-Provider LLM Factory
- Human-in-the-Loop Execution & Approval Workflow System
- Multi-Agent Orchestration & Reflection Agent
- Lightweight Observability System
- Repository Intelligence
- Redis Migration for Pub/Sub & Shared State
- Autonomous Project Builder with Task Graph
- Logs / Traces / Terminal Bottom Panel (Phase 13.5)
- GitHub Integration (as MCP Client - Phase 13.6)
- In-Editor Code Editing with Monaco (Phase 13.7)
- MCP Tool Protocol Exposing Core Tools as Servers (Phase 14)
- Local ML Layer (Phase 14.5)
- PostgreSQL Migration (Phase 15)
- Docker & Kubernetes Full Deployment (Phase 16)
Devraj Singh
Portfolio: https://devraj-singh.vercel.app/
GitHub: https://github.com/Devrajji-Singh
LinkedIn: https://linkedin.com/in/devraj-s/
MIT License