Quick Start

This guide takes you from nothing to your first AI response in about five minutes. You will deploy Grengin into your own cloud (or onto your own machine), run a short setup wizard, sign in, and send your first prompt.

Grengin ships as a pre-built cloud image with PostgreSQL, Nginx, and SSL tooling already installed, so most of the setup is automated and the wizard itself finishes in under two minutes.

Before you start

You will need:

Somewhere to run it. An AWS account for the fastest path (more clouds are coming), or Docker on any machine if you just want to try it locally.
At least one AI provider key. An OpenAI or Anthropic API key, so your team can start chatting. You bring your own keys, and when you self-host, Grengin never routes your prompts through its own servers.
A way to sign in. A Google or Microsoft (Azure AD) account to use for single sign-on. You can register your own OAuth app, or use Grengin’s hosted SSO proxy to skip that step.

You do not need a DevOps team, and you do not need your own domain. Every instance qualifies for a free *.grengin.com subdomain, which you can claim during setup.

Step 1: Deploy Grengin

Fastest: deploy to your cloud

Go to grengin.com/deploy and launch a pre-built Grengin image in your own cloud account. PostgreSQL, Nginx, Certbot, and the Grengin services come already installed and configured.

Grengin currently supports AWS, with more clouds on the way. Choose the available cloud option on the deploy page to get started.

On first boot, the setup wizard launches automatically at http://your-server-ip. Open that address in your browser to continue.

Or run it locally with Docker

To try Grengin on your own machine without a cloud server, run the installer image:

docker pull ghcr.io/perter-tech/grengin-image:latest
docker run --rm -p 80:80 ghcr.io/perter-tech/grengin-image:latest

Then open http://localhost in your browser. For a production self-host using Native (systemd) or Docker Compose mode, see the Installation guide.

Step 2: Complete the setup wizard

The wizard walks you through six steps. On a pre-built image, the first three are mostly automatic.

System check. Confirms your operating system, available ports (80, 443, and 8080), memory, and disk.

Database. Choose “Install PostgreSQL locally” for the quickest setup, or connect an existing PostgreSQL 14+ instance.

Domain and SSL. Enter your domain and get a free Let’s Encrypt certificate, or claim a free *.grengin.com subdomain. For a quick local test, a self-signed certificate also works.

Sign-in (OAuth). Connect Google or Microsoft (Azure AD) sign-in. To avoid registering your own OAuth app, use Grengin’s hosted SSO proxy instead.

AI providers. Paste your OpenAI or Anthropic API key. Keys are encrypted at rest with AES-256-GCM before they are stored.

Finalize. Grengin generates its secrets, runs database migrations, starts the services, and runs a health check. When it finishes, your instance is live.

Steps 4 and 5 are technically optional and can be set later from the admin panel, but you need at least one sign-in provider to log in and at least one model key to chat. Setting both here is the fastest route to a first prompt.

Open your instance (your domain or your free *.grengin.com subdomain) and sign in with the provider you just configured. The first user to sign in automatically becomes the admin, with access to the full admin dashboard. After that, new users are provisioned automatically the first time they sign in.

Step 4: Send your first prompt

From the chat screen:

Start a new conversation.
Pick a model from the model selector (for example, GPT-4o or Claude).
Type a message and send it.

Responses stream back in real time. From here your team can attach files, toggle web search per message, and save prompts to a personal library.

That is the full path from deploy to first response.

Add more models (optional)

The wizard connects OpenAI and Anthropic. To add more providers or manage keys centrally, open the admin dashboard and go to AI Engines, where you can add and validate keys for OpenAI, Anthropic, Mistral, and Gemini, then enable specific models. Keys can be set globally or per organization.

Next steps

Your instance is now running with governance on from the start. Common things to do next:

Invite your team. Add users and assign roles from the admin dashboard.
Set budgets and roles. Apply spend caps and role-based permissions by department or project.
Connect your tools. Add Jira, Confluence, Google Drive, or any MCP server.

Continue with:

Compare deployment modes and set up production SSL in Deployment.
Configure SSO, environment variables, and budgets in Configuration.