Evan Robertson

Fix everything — including legacy code

2026-10-01

The age of AI is here and we can fix everything. I've been happy with recent model launches since we received Opus 4. I thought it was fantastic. Opus 5.5 opened my eyes to a whole new world of possibilities. Suddenly, I have the time and resources to fix everything.

In a legacy codebase I work in frequently, there are issues everywhere I know of but have never had the time nor priority to fix. They were problems, but not the biggest fish to fry. Performance, security, clean code. I dreamed up fixes, wrote specs and prototypes, but actually implementing any of it in a codebase the size of this one wasn't feasible.

Until it clicked. I can take my extensive knowledge of Rust, ABL, GitOps, and our infrastructure, and just fix it. There is nobody telling me I'm not allowed to fix it, I have the ability to host and test everything myself, and Opus can implement anything I can. So why not?

That's what I'm up to these days. Taking a legacy codebase, typically deployed on-premises, and bringing it into the cloud, a Kubernetes cluster. I'm giving it new native libraries to step over shortcomings of ABL. I've set up true CI/CD, so regressions are caught early, and deployments are easy. Automated releases keep the version of each container nice and organised. And I'm finally being the difference I wanted to see.

No blackpilling, we're fixing everything.