Can AI help with Infrastructure Productivity

This post is the result of some notes I took whilst joining a live learning session from O’Reilly. One of the really interesting points made was that if you don’t want to be left behind as a result of AI, or perhaps you don’t want to lose your job to an AI agent, service, etc. Then you need to be using AI.

I think that is a valuable point, the AI stuff has come along loads in just a very short period of time, but if you’ve used it, you will likely agree that whilst there are some amazing things, there is also a lot of junk. In fact the real value comes from being able to work with an AI agent to get the best answers.

Infrastructure in the DevOps world

First of all, we’re not talking about the physical world here, we’re talking about the kinds of virtual infrastructure you have in Cloud environments etc.

GitHub Copilot

GitHub Copilot is like a pair programmer. It starts out not really knowing what you might need but as you work through your activities, it will start offering help.

Using GitHub Copilot from within VSCode, allows us to do things like create an AWS EC2 instance. Using useful file-names will help prime the pair programmer. Example:

Dir = aws-ec2 file= aws.create.sh

You start to wirte the script and it will make suggestions, the more you accept, the more confident the pair programmer gets in the suggestions it is making.

Pieces for Developers

Pieces monitors your actions and helps to make suggestions, it is basically the same as Copilot, but with come context as well.

K8sGPT

This is an agent that can help monitor your Kubernetes clusters

GitHub Copilot Workspace

Currently a beta product, more than a chat agent, it is an actual AI agent.

it allows you to work in a conversational style. It can break-down a problem into smaller parts and then complete those parts for you.

That is not to say that it can create your infrastructure, but it can brainstorm the idea with you and then help you build out the GitHub repository for you.

Is GenAI going to replace DevOps?

Personally I don’t think so. There are indeed common tasks in almost all jobs, these common tasks can be done quicker, with some good prompt engineering, byt an AI agent.

Therefore the benefit of using GenAI is to speed up the busy work so that you can shift your focus to the bigger picture items.

The less time you spend doing the boring things is more time you can work on up-skilling and it is more time you have available for your employer, essentially you’re saving the employer money, and talking about this on a CV might be the thing that gets you an interview for example.