With respect to prerequisites from end-user perspective -

Scenario:
When our end user login to intelops UI, the first thing we ask the user is to choose which cloud the User wants to use to set up the Control-Plane we provide & manage for them to set up air-gap control-plane behind the user's network firewall.
So we need to ask user to create a user id in the cloud IAM with full admin rights and we take those credentials (access key, secret key, tokens, password, etc. whatever is relevant to the cloud) in our UI's inputs form.
Using those credentials, we should create TalOS cluster with production-level resources configurations for cpu & memory & storage.
After the cluster is up, going forward we should use the TalOS cluster's API to connect to it & manage it.
So the first piece to deploy automatically as soon as the cluster is ready should be our gRPC agent which establishes the connection between our SaaS side and the control-plane we auto-created for customer inside customer's cloud. From there onwards we should use the TalOS API & gRPC agent to deploy & configure everything we support , into that TalOS cluster.

As first cloud we support we can start with assuming customers have AWS account.

Pramod - work this full automation & setup of TalOS for AWS.

https://github.com/replicatedhq/kURL