Transition to AI / Feeling left behind by agentic workflows
I'm a software engineer with roughly 7 years of experience, primarily writing backend C++ code for the aviation industry. Defense and aerospace have felt somewhat insulated from the broader shift, but we have tools like Copilot integrated into Visual Studio, which has been undeniably convenient.
I’ve been treating it almost like a 3D printer for code, using it to generate helper tools, widgets, or unit tests. However, I’m hesitant to let go of the reins. Once generated code reaches a certain complexity, I take a step back because I no longer fully understand every line, which quickly gets overwhelming.
At the same time, I see SWEs online talking about cranking out agentic AI workflows to fully automate complex tasks. Compared to that, I feel like I'm still in the stone ages cobbling tools together by hand while the industry shifts to mass production.
For those in a similar backend/embedded spaces, how much of this "agentic engineer" trend matches your day-to-day reality versus social media hype? If you’ve successfully integrated autonomous tools into your workflow without losing control of your codebase, what tools or practices made the biggest difference?