Skip to main content

Setup Wizard

Once you open http://localhost:8888 (or http://<server-ip>:8888), the wizard walks you through 9 short steps. Your choices are saved at every step, so you can close the browser and reopen http://localhost:8888 to resume exactly where you left off.

:::tip Read this section before clicking through Each step below describes what you'll see, what each option means, and any gotchas. The wizard's own helper text is intentionally short — this page is the long version. :::

1. Deployment Type​

What you do: Pick where Olympus will live. This is the most important choice — it gates which SSL options, DNS checks, and domain helpers appear in later steps.

Your options:

  • Public Network — internet-facing server. Unlocks Let's Encrypt auto-renewing certificates and real DNS validation. Best for production deployments on a cloud VM with a public domain you control.
  • Private Network — intranet, VPN, or air-gapped LAN. Use self-signed or custom certificates; DNS validation can be skipped.
  • Local Machine — laptop, dev box, single-user install. Auto-fills the domain as olympus.local and walks you through /etc/hosts setup. Skips DNS entirely.

:::info Heads up You can't switch deployment type later without redeploying. Public Network requires a real domain you actually control and the ability to point its DNS at this machine. :::

2. Domain Configuration​

What you do: Enter the primary domain Olympus will be reachable at (e.g. files.acme.com). All subdomains the platform needs — api., admin., mail., etc. — derive from this single value.

For Local Machine, you can type a short base name like olympus and the wizard turns it into olympus.local. You'll see the exact /etc/hosts lines you need to paste.

For Public / Private Network, the wizard validates that DNS records for your domain (and key subdomains) resolve to this server's IP. There's a Skip DNS check escape hatch for private networks where the DNS is internal-only.

Your options:

  • Primary domain (required)
  • Domain mode toggle (for Local Machine: short name → .local vs. custom full domain)
  • DNS validation pass/skip

:::caution DNS must propagate Even after you create the right A records at your registrar, propagation can take minutes-to-hours. If DNS validation fails, wait and retry — the wizard preserves your inputs. :::

3. Security & SSL​

What you do: Choose how the platform terminates HTTPS. The available options depend on the deployment type you picked in Step 1.

Your options:

  • Let's Encrypt (Public Network only) — free, auto-renewing certificates. The wizard asks for a notification email Let's Encrypt will send expiry warnings to. Requires ports 80 and 443 to be reachable from the internet during issuance.
  • Self-Signed (Private / Local only) — wizard generates a 3-year RSA-2048 cert with wildcard SANs covering all the platform subdomains. You can download the root cert to install on client devices so browsers don't warn.
  • Custom Certificates — upload your own PEM-encoded private key + certificate chain (leaf + intermediates). The wizard validates the chain, domain match, and expiry (must have ≥1 month remaining).

:::caution Public Network can't use self-signed The Let's Encrypt issuance flow requires Public Network mode. If you're on a private network but want a real certificate, generate one externally and upload via Custom Certificates. :::

4. Account Setup​

What you do: Set the initial administrator credentials and the internal PostgreSQL password the platform will use to talk to its own database.

Your options:

  • Administrator username (defaults to admin, editable)
  • Administrator email — used for license activation in a later step and password recovery
  • Administrator password — has a Generate button for a secure random value
  • PostgreSQL password — also has a Generate button

:::caution Avoid these characters in passwords The wizard rejects #, @, :, /, ?, %, [, and ] — these characters break URL-encoded database connection strings. The Generate button produces a safe value automatically. :::

5. GenAI Settings​

What you do: Wire up AI providers so the platform is fully functional the moment it boots. Everything on this screen is optional — the platform deploys cleanly with all AI providers disabled, you can just configure them later in the Admin Panel.

Public providers (work everywhere):

  • Google Gemini — API key + model selection
  • OpenAI — API key + model selection
  • Anthropic Claude — API key + model selection

For each provider, the wizard validates the key and shows you the catalog of models it can actually see.

Self-hosted providers (NVIDIA GPU required):

  • Ollama — local LLMs. Disabled / Self-Hosted (with model pull list) / Remote URL (point at an existing Ollama server)
  • Stable Diffusion — local image generation. Couples with self-hosted Ollama, requires a GPU

Default model roles — at the bottom, pick which configured provider+model handles each role: embedding, chat, summary, vision, MCP chat, MCP image.

:::info No GPU? Skip the self-hosted section Self-hosted Ollama and Stable Diffusion are auto-disabled on hosts without an NVIDIA GPU. Public providers work fine without one. The platform also deploys cleanly with everything disabled — you just won't have AI features until you configure them later. :::

6. License​

What you do: Choose your edition. Olympus comes in two flavors.

Your options:

  • Olympus Core (Community) — free forever, no license key needed. Just select and proceed.
  • Olympus Business & Enterprise — paid edition with two sub-paths:
    • I have a license — paste your key in the OLY-LIC-XXXX-XXXX-XXXX-XXXX format. The wizard validates it and shows max users, expiry, and the device fingerprint it'll be locked to.
    • Start free trial — name + email gets you 30 days of all Enterprise features, up to 10 users.

A feature-comparison table on the screen shows exactly what each edition includes.

:::caution Enterprise licenses are device-locked The trial and paid license bind to this machine's hardware fingerprint (OS, CPU, hostname). Moving the deployment to a new server requires regenerating the license — contact support before doing that. :::

7. Configuration Preview​

What you do: Read-only review of the actual config files the wizard will generate during deployment — .env, appConfig.json, production.js, init.sql, the Docker Compose files, and the nginx vhost.

A file list runs down the left side; click a file to see its contents with syntax highlighting on the right. Scan for the values that matter most: the right domain, the right SSL mode, the AI providers you actually enabled.

Heads up: Auto-generated security keys (JWT secrets, VAPID keys, etc.) are not shown here — they're generated at deployment time, not at preview time. Don't worry that you can't see them.

8. Optional Services​

What you do: Toggle on optional sidecar services. All four sections default to off and the platform deploys fine without any of them.

Your options:

  • MailHog (local email testing) — catches all outbound email and exposes it at mail.{domain}. Useful for dev and demos, never enable in production — real customers won't get emails.
  • File Sharing (NFS / SMB) — create up to 3 containerized file servers that mount host paths into the platform. NFS isn't available on Windows hosts; use SMB there.
  • Primary Users — only appears if you set up at least one NFS mount. Create login accounts that get immediate access to the share when the platform boots.
  • OpenClaw Bot (Enterprise only) — multi-tenant AI assistant. Toggle to enable.

:::caution File-sharing paths are not auto-created The wizard format-validates the host paths but doesn't check they exist. Make sure the directories are created on the host before deploying, or the mounts will fail at container start. :::

9. Final Deployment​

What you do: One last full review — a 6-section summary showing deployment type, domain, SSL mode, admin account, license edition, AI providers, and optional-services counts.

Below the summary, a checklist shows exactly what the deployment will do: generate configs, start ~29 containers, initialize the database, request/install SSL, bring the platform online.

Tick the "I have reviewed everything" box, click Deploy & Start Platform, and watch a live progress feed. Deployment takes 2–5 minutes depending on whether images need to be pulled.

:::caution One-way door After clicking Deploy, you can't edit prior steps in this wizard run. If something looks wrong on the summary, fix it now. (You can always re-run the wizard later for a clean redeploy.) :::

When the progress feed turns all green, the wizard auto-redirects you to your new Olympus deployment. Continue to Post-Install → to verify everything is healthy.