Your own server, not a row in someone else's table
Most AI builders run every app on shared infrastructure, in a shared database, within reach of the same agent that wrote it. A cloud project here gets a dedicated server by default (or joins one you already run) with PostgreSQL installed on that server, a domain, TLS, and a way to leave.
What runs where
- One server per project by default. The Builder provisions a cloud VPS for the project and installs PostgreSQL 16 on it. You can also place more than one of your projects on a server you already have; each keeps its own database, hostname and process, and the cap is shown when you provision. Nothing is shared with another tenant.
- A normal Linux service. The app runs under systemd as
proj-{slug}; Caddy sits in front and gets a certificate from Let's Encrypt automatically. - Deploys over SSH. When you redeploy, the Builder sends the project to the server, installs dependencies, builds it and restarts the service.
- Server backups. Automatic server backups are enabled for provisioned servers.
Regions and sizes
Read live from the same settings the product bills from. Servers are a flat monthly price on top of a paid plan.
| Region | Size | vCPU | Memory | Disk | Per month |
|---|---|---|---|---|---|
| Europe (nbg1) | small | 4 | 8 GB | 80 GB | $20 |
| Europe (nbg1) | medium | 8 | 16 GB | 160 GB | $40 |
| Europe (nbg1) | large | 16 | 32 GB | 320 GB | $75 |
| United States (ash) | small | 4 | 8 GB | 160 GB | $45 |
| United States (ash) | medium | 8 | 16 GB | 240 GB | $80 |
| United States (ash) | large | 16 | 32 GB | 360 GB | $150 |
Domains and TLS
Every cloud project gets a *.cloud.seamai.pro hostname. Register a domain from inside the Builder or point one you already own at the server; Caddy provisions the certificate from Let's Encrypt and renews it.
Leaving: export and self-host
The project is a standard layout with no proprietary runtime. Download it as a ZIP or push it to a GitHub repository, then run it anywhere:
- Export: Download as ZIP from the workspace, or push to GitHub with your own token.
- On your machine or server:
npm install. - Set the environment your app expects (for cloud projects,
DATABASE_URLpointing at a PostgreSQL you control). - Build with the project's own script:
npm run build. - Run the built server under systemd, PM2 or a container, with a reverse proxy for TLS.
The docs walk through it in more detail.
What this costs, plainly
Cloud servers are available on paid plans and billed flat per server size; the sizes and prices above are the real ones. AI usage beyond your plan bills per 1K tokens. See pricing.