Google AI Studio can generate a working application in minutes, but not one that reaches a database behind your firewall. This walkthrough keeps every port closed: DreamFactory exposes the SQL Server as a REST API governed by a read-only role and a scoped API key, and AI Studio builds its delivery operations dashboard against that API instead of against the database.
What you’ll learn
- Why Google AI Studio has no route into a private network
- Exposing on-prem SQL Server as a governed REST API with no open ports
- What the read-only role and scoped key actually restrict
- Building a working dashboard in AI Studio on live on-prem data
Frequently asked questions
Can Google AI Studio connect to a private or on-premise database?
Not directly. AI Studio runs in Google's cloud and cannot reach into your network. What it can do is call an HTTPS endpoint, so putting a REST API in front of the database — as DreamFactory does in this video — is what makes the data reachable.
Do I need to expose SQL Server to the internet for AI Studio?
No, and that is the point of the build. The SQL Server accepts no inbound connections from outside. DreamFactory holds the connection and publishes a governed REST API, so the only thing reachable from the internet is that API.
How is access limited once the API exists?
By role. The API key AI Studio uses is bound to a read-only role scoped to specific tables, so the generated application can read what it needs and nothing else — regardless of what the model decides to request.
