GT AI OS Community Edition

GT AI OS Community Edition is a self-hosted, web-based generative AI platform for individuals and teams who need document-centric workflows with strong data-privacy controls. Install on Ubuntu (x86_64), NVIDIA DGX OS 7 (ARM64), or Apple Silicon macOS using Docker and the runbooks in this repository’s wiki.

🚀 New: GT AI OS Gen 3 is here. Check out our next-generation platform at github.com/GT-Edge-AI/GT-AI-OS.

---

Installation

Choose your platform for step-by-step instructions:

Each platform uses its own install script. Select the guide that matches your OS and CPU architecture.

Start here: Installation (wiki)

---

Update an existing installation

From your clone of this repository:

macOS:

cd ~/gt-ai-os-community && git pull && bash scripts/deploy.sh

Ubuntu:

cd ~/gt-ai-os-community && git pull && bash scripts/deploy.sh

DGX:

cd ~/gt-ai-os-community && sudo git pull && sudo bash scripts/deploy.sh

For troubleshooting and release notes, see Updating.

---

Access

Change default passwords after first sign-in in production use.

---

Platform requirements

A typical install uses about 7 GB RAM at steady state. Local models, conversation history, and datasets require additional disk space.

Supported: Ubuntu on Proxmox with GPU passthrough. Not supported: Windows hosts.

Install scripts for macOS target Apple Silicon only (not Intel Macs).

---

Inference and privacy

• Connect local inference (Ollama) or external APIs (NVIDIA NIM, Groq, vLLM, SGLang, and others supported in the Control Panel).
• For document workflows, use local models or providers that offer zero data retention when privacy is required.
• As of v2.0.33, Community Edition is not multimodal: it does not generate or process images, video, or audio in the core product path.

---

Embeddings and GPU acceleration

Retrieval-augmented generation (RAG) uses an embedding model to index uploaded files. NVIDIA GPUs and Apple Silicon accelerate embedding and dataset ingestion.

---

Features

• Agent builder — Custom agents with system prompts, categories, role-based access, and guardrails
• Local models — Run models with Ollama for offline inference
• Document processing — Datasets and RAG-backed chat over your files
• Teams — Shared access to agents and datasets within a workgroup
• Observability — Usage dashboards, chat logs, and operational metrics

---

Documentation

Full guides are in the wiki:

---

Quick commands

docker compose ps              # Service status
docker compose logs -f         # Follow logs
docker compose down            # Stop stack
docker compose up -d           # Start stack

---

Community vs Enterprise

Enterprise: Contact GT Edge AI

---

Architecture

┌────────────────────────────────────────────────────────────────┐
│                          GT AI OS                              │
├──────────────────┬──────────────────────┬──────────────────────┤
│   Control Panel  │      Tenant App      │   Resource Cluster   │
│    (Admin UI)    │       (User UI)      │ (AI inference routing)│
├──────────────────┴──────────────────────┴──────────────────────┤
│                         PostgreSQL                              │
│                  Control DB  │  Tenant DB                       │
└────────────────────────────────────────────────────────────────┘

---

Support

• Runbook issues: GitHub Issues
• Security: SECURITY.md and contact GT Edge AI

---

License

Apache License 2.0 — see LICENSE.

---

GT AI OS Community Edition · GT Edge AI