Login with your deployment credentials
Navigate to deploy.langsmart.app/wizard. You'll be greeted with a login screen. Use the username and password provided by your Aperion account manager — these are tied to your organisation and license.
Enter your server IP and hostname
After login, the wizard shows your organisation name and license type. Enter the public IP address of the server you want to deploy Smartflow on, along with the desired public hostname (e.g. ai.yourcompany.com). This hostname will be used to configure TLS and generate the Caddyfile.
Provide SSH access to your server
The wizard needs SSH access to install Docker, clone configuration, and run Compose. You can provide either an SSH private key (paste it directly) or a root password. Credentials are used only for this session and are never stored.
SSH Private Key
Paste your private key
Password
Root password
MIIEowIBAAKCAQ...
Test the SSH connection
Click Test Connection to verify the wizard can reach your server. The backend attempts an SSH handshake and returns the server OS, RAM, disk, and Docker status within a few seconds. Fix any issues before proceeding.
ubuntu or admin if root fails), or key format mismatch (must be PEM, not OpenSSH format). Click Test Again after fixing.Choose your deployment type
Select how you want to run Smartflow — a single Docker Compose stack for most deployments, or Kubernetes for teams that already run their own cluster. Both options produce a production-ready configuration.
Single-Server Docker Compose Recommended
All Smartflow services run on one server via Docker Compose. Best for most teams. Up and running in under 5 minutes.
Kubernetes (Bring Your Own Cluster)
Deploy to your existing K8s cluster using our Helm chart. Requires kubectl access and a valid kubeconfig.
Configure your environment
Enter your AI provider API keys, set your admin credentials, and configure optional features. Only the keys for providers you plan to use are required. All values are written to a .env file on your server — never stored on Aperion infrastructure.
Deployment — watch it happen live
The wizard SSHes into your server and runs the deployment. Every command and its output streams live to your browser via Server-Sent Events. You can watch Docker pull images, start containers, and run health checks in real time. This typically takes 3–6 minutes depending on your server's internet connection.
Test query and go live
Once all containers pass health checks, the wizard runs a live test query through your new proxy and displays the response alongside the full VAS trace log. A popup shows your initial admin credentials. From there, you're redirected to your new Smartflow dashboard.
https://ai.acmecorp.com/dashboardmy-secure-admin-key-2026Kubernetes deployment path
If you selected Kubernetes at step 5, the wizard collects your kubeconfig (paste or upload), sets up the smartflow namespace, and applies the full Helm chart. The same live log stream shows pod status as they reach Running.
Paste your kubeconfig
The wizard accepts a standard kubeconfig. Only the cluster you choose is used — it does not read other contexts. The kubeconfig is discarded at session end.
Namespace and secrets are auto-created
The wizard creates the smartflow namespace and populates the smartflow-secrets Secret with your API keys and admin credentials.
Helm chart applied, pods watched live
All Smartflow services are deployed via Helm. Pod status streams live — you see each pod move from Pending → Running → Ready.