Software Developers in the age of AI

At the end of 2024 I’d just started using LLMs such as ChatGPT and Gemini for rewriting content – copy/pasting my existing content and prompting it to improve it in certain ways or adapting content to new designs. A lot of the Version 2.0 tutorials on this site were even migrated using this method along with some manual review.

Then 2025 came and went and during that year I’d 1) started using AI agents (Copilot) in Visual Studio and Visual Studio Code, and 2) lost a mid-level developer from my team, leaving only myself to extend and maintain the warehousing software I wrote.

We had a moment there where we deliberately didn’t hire anyone because we wanted to assess what type of development we wanted to pursue – more C# development? someone cross-functional who could also work on our React website? someone purely to work on the React website?

The answer ended up being, we didn’t hire anyone… because we didn’t need to.

I used this as an opportunity to go nuts on AI agents, creating several folders for our software, each on a different branch to work on a different ticket. Instead of managing people, I was managing AI – about 5-6 at a time before I couldn’t keep up with what each one was doing.

I’ve reduced our backlog of tickets from 120+ tickets to about 70 as of writing. This may not sound like much but is huge as we hadn’t been sub-100 for years. Little menial tickets we didn’t have time for – the AI had time for while I got on with bigger tasks or architecting.

This doesn’t mean you can just let the AI agents go nuts and call it a day. You need to prompt them (they are surprisingly resilient to poor prompts), you need to review their code, you need to test it as you would your normal work. The slop is real and it takes an experienced programmer to enforce logical software layers, coding standards and even prompt them to follow basic practices like SOLID or even DRY.

Not only is the slop real but they also have a habit of running PowerShell commands that simply aren’t available, over and over again, while also truncating files randomly. The agents also struggle a LOT with files over 1500 lines of code, they truncate those more than anything and I often work with them to split large legacy files into smaller bite-size files just so that the AI can work with them.

They (the AI agents) need to be managed, just like people.

Back to the people side of things though, I’ve been discussing with friends in the IT space and it’s a crazy time. Computers are coming for our jobs and as a senior developer I no longer need to hire junior or mid-level developers – they’ve already been replaced.

I know that sounds dramatic – but I’m not saying this as some “AI hype” person. I’m saying it because I literally lived it this year. We had a dev leave, we didn’t replace them, and we still shipped stuff and knocked over a pile of backlog.

The AI space will continue to improve and put people out of jobs. It starts with those of us that sit behind a desk, but we see it heavily influencing the robotics space and it really won’t be long until physical jobs will be at risk as well.

What’s the future of my role look like? In the future I will no longer be an IT lead or senior software developer, I’ll probably end up being a business analyst and AI prompter. My skills during that will be invaluable in keeping the AI in check but that won’t last forever.

The weird part is I don’t even know if that’s a good thing or a bad thing yet. It’s just… what happened.