Love it! Coincidentally, https://infrastructurefromcode.com/ . The key piece is a generative AI ”CoPilot” built atop a foundational LLM model and fine-tuned for public cloud infrastructure concepts and end points. Terraform-generation (Open Tofu) maybe a first step, similar to how GitHub CoPilot can generate Java, etc. code.
Great insights! I used ChatGPT to debug and resolve deployment errors. I paste logs, it detects the issue, and provides a clear, step-by-step resolution guide.
As a way to more easily convert between cloud providers, yes I can see that. Whether, especially from a security perspective, auto-generation from some other inferred abstraction, I’m yet to be convinced (perhaps because I’ve been around long enough to hear that promise repeated over and over and yet never quite delivered). I am minded by the emphasis Microsoft place on the word ‘Co’.
I think I would also take issue with the assertion that writing or at least providing context configuration by app team who don’t operate as a DevSecOps team is ‘hard’. Writing templates from scratch in an unfamiliar DSL, yeah ok. But providing a few values in a tfvars.yml or a small overlay template ? Yeah one or two members of the may have to have slightly more understanding, and yes, it is a constraint model, but, ‘hard’, not really.
Shift-left means different things to different orgs these days, but TBH I think there are bigger ‘fish to fry’ for AI than IaC, and if that’s all its going to offer, then its a big fuss about not very much (I appreciate you are not saying that, I’m just trying to suggest that AI maybe better employed on helping with some harder problems than this one).
Love it! Coincidentally, https://infrastructurefromcode.com/ . The key piece is a generative AI ”CoPilot” built atop a foundational LLM model and fine-tuned for public cloud infrastructure concepts and end points. Terraform-generation (Open Tofu) maybe a first step, similar to how GitHub CoPilot can generate Java, etc. code.
Hey thanks a lot for sharing this!
Great aid
Great insights! I used ChatGPT to debug and resolve deployment errors. I paste logs, it detects the issue, and provides a clear, step-by-step resolution guide.
As a way to more easily convert between cloud providers, yes I can see that. Whether, especially from a security perspective, auto-generation from some other inferred abstraction, I’m yet to be convinced (perhaps because I’ve been around long enough to hear that promise repeated over and over and yet never quite delivered). I am minded by the emphasis Microsoft place on the word ‘Co’.
I think I would also take issue with the assertion that writing or at least providing context configuration by app team who don’t operate as a DevSecOps team is ‘hard’. Writing templates from scratch in an unfamiliar DSL, yeah ok. But providing a few values in a tfvars.yml or a small overlay template ? Yeah one or two members of the may have to have slightly more understanding, and yes, it is a constraint model, but, ‘hard’, not really.
Shift-left means different things to different orgs these days, but TBH I think there are bigger ‘fish to fry’ for AI than IaC, and if that’s all its going to offer, then its a big fuss about not very much (I appreciate you are not saying that, I’m just trying to suggest that AI maybe better employed on helping with some harder problems than this one).