Provision Cloud Machines on Demand with the /v1/machines API
Most automation stacks rely on static selectors or a limited set of APIs. When a UI changes or an app adds a native dialog, your scripts break. The /v1/machines API lets you provision live cloud machines on demand. You can then drive those desktops with a computer use agent that sees the screen and acts like a human. This gives you OS-wide automation, browser workflows, and terminal tasks without brittle IDs.
How it works
The workflow starts with a POST to /v1/machines. You specify a machine_id (your own identifier) and a cloud provider (aws, azure, gcp). The server provisions a VM and returns a status. Once the status is "running", you can start sending predict calls to that machine via the session API. The agent interacts with the full OS, browser, or terminal using vision and mouse/keyboard actions. You can stop the machine or take a snapshot when you are done.
curl -X POST https://coasty.ai/v1/machines \
-H "X-API-Key: $COASTY_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"machine_id": "my-prod-vm-01",
"provider": "aws",
"size": "t3.medium"
}'Request fields
- ●machine_id: a unique string you own for tracking (e.g., my-prod-vm-01).
- ●provider: one of the supported cloud providers (aws, azure, gcp).
- ●size: VM size configuration (e.g., t3.medium).
Response fields
- ●machine_id: the identifier you submitted.
- ●provider: the cloud provider the server chose.
- ●status: current state (e.g., provisioning, running, stopped).
- ●region: cloud region used for the VM.
- ●ip_address: public IP of the provisioned machine.
- ●created_at: ISO timestamp when the machine was created.
- ●updated_at: ISO timestamp of the last status update.
POST /v1/machines with machine_id, provider, and size to spin up a cloud VM. Check the status field to know when it is ready to drive.
Where this beats brittle automation
Traditional automation tools rely on selectors, IDs, or documented APIs. When a UI changes or a vendor adds a native dialog, those selectors stop working. A computer use agent that sees the screen can locate elements by description, understand context, and use natural language instructions. It can handle drag-and-drop, native dialogs, and browser popups without brittle selectors. Pairing /v1/machines with a computer use agent gives you robust automation across browsers, desktop apps, and terminals.
Start building agents that operate on real machines. Create workflows that spin up VMs, drive them through tasks, and clean up afterward. Get a key at https://coasty.ai/developers and try the API today.