Post

I think that how accurate this is depends on the type of software you're thinking of. I think AI will help for all software. But if your problem is that you have a ton of existing data at rest, unpredictable workflows, user transactions that hit multiple services, etc, AI gets you less leverage.

Specifically, I am thinking of these as issues that raise the cost of answering "is this change safe?" in a context where you don't have the evidence that conclusively proves/disproves safety.

I spent (far too much) time optimizing my vibe-coded c++ compiler. I learned that agents in optimization loops are amazing at micro-optimization and absolutely terrible at macro-optimization. Weeks of optimizing with a lot of guidance was barely able to get within an order of magnitude of gcc/clang.

The only way I was able to actually get there was to have the agent rewrite from scratch focusing on perf from the beginning. With a good foundation, 20 hours or so of micro-optimization in a loop got compile times down to state of the art.