I am considering using the automation api to manage resources I am provisioning on behalf of my users. I have pulumi stack code written and parameterized in the way I need. The thing is, my application is a JVM app. As such, I’m pondering how to get this whole setup working in a prod scenario. My current thinking is to bundle my pulumi stack code into a single js executable (use webpack or ncc), package that into an executable binary (e.g., via pkg), add that binary to my container, and, finally, add the pulumi cli to my container. I can then execute pulumi stack operations by interactive with the binary executable I produced. Does this sound like a typical workflow for an automation api user?

That would work. Your Pulumi code is using Node.js?

Agree that it will certainly work. Sounds like you could potentially benefit from Pulumi Deployments - it's basically a managed service wrapped around automation API. It provides REST APIs for running pulumi programs on demand. https://www.pulumi.com/docs/pulumi-cloud/deployments/

Yep, it’s currently written in typescript. It feels like a very roundabout way to approach it, but it does seem like it’d work. A POC of the approach is functional. And yeah @lemon-agent-27707, I saw. It looks like a good fit, but I had some reservations. I really wanted this service to be designed as simply as possible, keeping dependencies to a minimum. Perhaps a bit silly since it already has deep coupling to AWS. The lack of pricing also gives me some uncertainty. Maybe I should more deeply evaluate deployments to avoid this whole bundling dance, haha.

Understand the concerns around unknown pricing. Pricing will likely match GitHub Actions. We have already started signing pricing agreements with a few customers who are cost-conscious and wanted to get something in writing. So always an option if that becomes a blocker for you. There's a #pulumi-deployments channel for discussion as well.