# Setup wizard

Answer five questions about your infrastructure and copy the `docker run` command for your CKEditor AI On-Premises deployment.

> **Note**
>
> The wizard covers the settings most deployments need. [Required configuration](configuration.md) lists every option.

<a id="prerequisites">

## Prerequisites

The generated command contains placeholders in square brackets. Collect these values first:

* **License key** – from the [Customer Portal](https://portal.ckeditor.com/), or [contact us](https://ckeditor.com/contact/) for a trial key.
* **Management secret** – a string you generate yourself. It grants access to the Management Panel, so keep it secret and reuse it across restarts. See [Secret keys](configuration.md#secret-keys).
* **Database host, user, password, and name** – an empty MySQL 8 or PostgreSQL 12 database that the container can reach.
* **In-memory data store host** – a Valkey or Redis instance that the container can reach.
* **LLM provider credentials** – one set for each provider you select in step 4.
* **File storage credentials** – for Amazon S3 or Azure Blob Storage. The filesystem and SQL database options need none.

<a id="setup-wizard">

## Step 1: Deployment mode

Standalone AI Run CKEditor AI On-Premises on its own. Select this if you do not run Collaboration Server On-Premises. AI + Collaboration Server Connect to [Collaboration Server On-Premises](../cs-onpremises/overview.md). The two products share one SQL database and one in-memory data store.

## Step 2: Database

MySQL 8+ PostgreSQL 12+

## Step 3: File storage

Amazon S3 Azure Blob Storage Filesystem SQL database

## Step 4: LLM provider

Select one or more providers. You can add more providers later.

OpenAI, Anthropic, and Google each have a default list of models. They work as soon as you add an API key.

Azure OpenAI, Amazon Bedrock, Google Vertex AI, and Custom have no default models. For these, declare the models in the `models` option. See [Custom models](llm-providers.md#custom-models).

OpenAI Anthropic Google (Gemini) Azure OpenAI Amazon Bedrock Google Vertex AI Custom (OpenAI-compatible)

Select at least one provider.

## Step 5: Optional features

MCP support Connect to external MCP servers. See the [MCP tools](mcp-tools.md). OpenTelemetry Export traces to an OTLP-compatible backend. See the [Observability guide](observability.md). Langfuse Export AI spans to Langfuse. Langfuse shows dashboards for models, tokens, and cost. See the [Observability guide](observability.md).

## Result

Copy the command. Replace each placeholder in square brackets with your own value. [Deployment](deployment.md) explains the steps around this command, and [Collaboration Server integration](collaboration-server-integration.md) explains the shared database and in-memory data store.

docker run

```bash
```

<a id="after-you-run-the-command">

## After you run the command

The container starts, but it serves no requests yet. Three steps remain:

1. [Create an Environment and Access Key](deployment.md#step-4-create-an-environment-and-access-key) in the Management Panel.
2. [Create the token endpoint](deployment.md#step-5-create-the-token-endpoint) in your application. It issues a JWT for each user. See the [Node.js example](../../examples/token-endpoints/nodejs.md).
3. Work through the [Quickstart checklist](quickstart-checklist.md). Its last step sends one request that uses your token endpoint, the access key, the database, and your LLM provider, so it proves that all four work.

---

Full index of the Cloud Services documentation: [llms.txt](../../../llms.txt)