Try it free

Documentation  /  Getting started

Installation

Docker Compose on your own hardware. One command brings the platform up, and the guide is tested by following it verbatim on a clean machine.

Before you start

  • A host meeting the sizing requirements
  • Docker and Docker Compose
  • An NVIDIA GPU with current drivers and the container toolkit, if Genie is running locally
  • A licence file

Install

01

Get the bundle

Download the release archive and verify its checksum against the published value. A bundle that does not match is not installed.

02

Configure

Copy .env.example to .env and set the database password, the base URL and the data directory. No default credential ships with the platform — the file will not start without values.

03

Bring it up

docker compose up -d. First start runs migrations and reports each one. It does not proceed past a failed migration.

04

Import your licence

Through Settings → Licence, or dg licence import ./licence.dgl.

05

Create the first account

The bootstrap routine issues a one-time credential printed to the container log and requires a change on first sign-in. There is no default password to remove afterwards.

06

Verify

dg health reports every service, the database, the model endpoint and the collector fleet. Anything unreachable is named rather than omitted.

GPU

Genie runs on your GPU. Confirm the container can see it before first use:

docker compose exec wayne nvidia-smi

If no GPU is present the platform runs and every deterministic module works. Genie's reasoning surfaces report model unavailable rather than degrading silently to something worse.

If the install guide is wrong

That is a defect. The guide is tested by following it verbatim on a clean machine, and a step that does not work as written fails our own build. Report it and we fix the guide, not just your install.

← Requirements and sizing  ·  Air-gapped operation →

Not running it yet? Start a thirty-day evaluation — read-only, on your own estate, no card.

Thirty days · read-only · no card

Run it beside what you already have, against your real clients. It tells you what your tools are reporting that is not true.

Start an evaluation