Installation

Install Full On-Prem with the email config pack and an NVIDIA GPU (online or air-gap).

Installation

Choose the path provided by Open Ticket AI.

This page covers Full On-Prem. One package supports OTOBO, Zammad, or KIX.

What you need (Full On-Prem)

• Docker Engine 24+, Docker Compose v2, and NVIDIA Container Toolkit
• NVIDIA GPU with ≥ 24 GB VRAM (nvidia-smi active)
• Config pack (otai.yml) from your email
• Ticket system API access (or air-gap release bundle)
• ≥ 32 GB host RAM

Config pack (from your email)

For pilots or licenses, we send:

1. otai.yml — version, ticket system, and env var key names for license JWT and optional hosted LLM API key.
2. The install script.

The installer generates API tokens and passwords automatically. Use the version specified in otai.yml.

Example otai.yml (default env keys):

version: 0.2.0
ticket_system: otobo # otobo | zammad | kix
license_env_key: OTAI_LICENSE_KEY
llm_api_key_env_key: OTAI_LLM_API_KEY

Set environment variables before installation:

export OTAI_LICENSE_KEY='eyJ...'           # JWT from your Open Ticket AI email
export OTAI_LLM_API_KEY='sk-or-...'        # only if you use a hosted LLM

$env:OTAI_LICENSE_KEY = 'eyJ...'
$env:OTAI_LLM_API_KEY = 'sk-or-...'

Do not paste keys directly into otai.yml.

Online install

Set env variables and run on the GPU host:

curl -fsSL https://download.openticketai.com/download.sh | bash -s -- --bundle online
cd otai-*   # extracted release directory
cp ../otai.yml ./otai.yml   # if file lived next to download
./install.sh

install.sh:

1. Verifies host RAM, free disk space, and NVIDIA GPU.
2. Generates .env, loads container images, and copies integration packages.
3. Starts Compose services.
4. Installs the otai CLI command.
5. Configures optional daily backups (enabled by default).

Non-interactive install

export OTAI_INSTALL_ASSUME_YES=1
export OTAI_ENABLE_BACKUPS=1          # default; set 0 to skip
export OTAI_BACKUP_SCHEDULE=daily     # daily | weekly
./install.sh

Air-gap install

1. Download the air-gap bundle on a connected host:

   export OTAI_VERSION=0.2.0   # matching otai.yml
   export OTAI_LICENSE_KEY='eyJ...'
   curl -fsSL https://download.openticketai.com/download.sh | bash -s -- --bundle airgap

2. Copy otai-<version>/ and otai.yml to the GPU server.

3. Run the installer locally:

   cd /path/to/otai-<version>
   cp /path/to/otai.yml ./otai.yml
   # optional: OTAI_OLLAMA_MODELS_DIR=/path/to/ollama/models
   ./install.sh

Containers validate licenses locally without network calls.

After install

1. Open Studio in your browser via the printed URL.

2. Install the OpenTicketAI add-on into OTOBO / Znuny (optional: OTAICustomerChat), or the frontend plugin for KIX.

3. Verify setup:

   otai verify
   otai version

If otai is missing, add ~/.local/bin to your PATH.

Host requirements (checked at install)

Full On-Prem requires an NVIDIA GPU. RAM or disk warnings do not block installation.

Recommended procurement configurations

Deployment requirements:

• GPU: NVIDIA GPU with ≥ 24 GB VRAM (e.g., RTX PRO 4000 Blackwell or NVIDIA L4)
• RAM: ≥ 32 GB recommended
• CPU: 4+ core x86-64 processor
• Storage: ≥ 250 GB SSD

2U rack forms, redundant PSUs, ECC RAM, and RAID 1 are optional resilience features. The passive NVIDIA L4 requires server airflow; workstation GPUs require verified chassis power and cooling.

Example hardware configurations (no commercial affiliation):

• Tower: MIFCOM workstation with RTX PRO 4000 Blackwell
• Rack with RTX PRO 4000 Blackwell: AIME A4004
• Rack with NVIDIA L4: Thomas-Krenn 2U AMD GPU server

Ensure 24 GB VRAM minimum (e.g., RTX 4000 Ada with 20 GB is insufficient).

Persistent data

Config stays in otai.yml, secrets in .otai/secrets.env. Compose .env is generated automatically. Ticket data remains in your ticket system.

Next steps

• Backups and restore
• Updates
• Handbook overview