Idea to deployed, without leaving the page
Describe what you want. Studio writes the code, runs the tests, creates the database it needs, and deploys it to a URL you can share. All in one place, on infrastructure you own.
No credit card · Stops when idle · Per-second billing
waiting for a prompt
The container it builds in is the container that ships
No dev and prod gap
Studio runs in your own private network on SMLL, not a sandbox that resembles one. What works while you are building it is what runs when you deploy, because it is the same environment.
Real infrastructure, not mocks
When your app needs a database, the agent creates a managed Postgres and wires the connection string in. Same for object storage, caches and queues. They are yours, and they outlive the conversation.
It picks the right target
A site that needs no server is uploaded to storage and served over HTTPS. Anything that needs a runtime becomes a container image in your private registry. You are not paying for a pod to serve files.
Four steps, and you are live
- Step / 01
Describe it
Plain language. Tag an existing database or bucket with @ and the agent uses it instead of making another.
- Step / 02
Watch it work
Every file written and command run is shown as it happens, along with what the tokens are costing.
- Step / 03
Check it
Open the editor, read the code, run the tests yourself. It is ordinary code in a container you control.
- Step / 04
Deploy
One action. Studio picks a container or static hosting, builds it, and hands you the URL.
Before you start
- What is SMLL Studio?
- An AI coding agent that runs in a real container inside your own private network on SMLL. You describe what you want in the browser, and it writes the files, runs the build and tests, provisions any database or storage it needs, and deploys the result to a live URL.
- How is this different from an AI code editor?
- An editor gives you code and leaves deployment to you. Studio develops in the same container that gets deployed, so there is no gap between what worked locally and what runs in production, and the infrastructure it creates is real managed infrastructure on your account rather than a mock.
- Does it deploy to a container or a static site?
- Whichever fits. Studio inspects the project: anything needing a server at runtime becomes a container image pushed to your private registry and run as a service, and a build that emits plain files is uploaded to object storage with static hosting. A static site costs storage rather than a running container.
- What does it cost?
- You pay for the tokens the agent uses and for the environment while it is awake, both at published per-unit prices. Studios stop automatically when idle, so an environment you are not using costs storage only. There is no separate subscription.
- Can I edit the code myself?
- Yes. Studio has a full editor and a terminal alongside the agent, and the files are ordinary files in a container you control. You can take over at any point, and everything the agent creates is standard code with no lock-in.
- Which AI models can I use?
- Any tool-capable model in the SMLL catalogue, switchable mid-conversation. Cheaper models handle routine edits and stronger ones handle harder problems, and the running cost is shown in the editor as you work.
Build the thing you keep meaning to
Open a studio, describe it, and have a URL to send someone before you have finished your coffee.
Free tier included · Stops when idle · No credit card
