There's no reason for software to be slow anymore

	There's no reason for software to be slow anymore(danluu.com)
	464 points by Jach 11 hours ago \| 328 comments
	tl;dr: LLM coding agents have dramatically lowered the cost of performance optimization work that previously required rare expertise, making it feasible to apply specialized techniques (JIT compilers, custom indexes, multi-threading) to problems where they wouldn't have been worth the effort. The author demonstrates this by having agents build workload-specific optimizations for ripgrep in minutes, and cites examples like an Azul game AI that beat academic SOTA with far less compute. This enables a shift toward "dynamic custom software" fitted to specific workloads rather than general classes, though agents still struggle with open-ended experimental design.
	HN Discussion: ↑Agents are highly effective at optimization work, confirming the article's thesis with personal experience ↓Slowness is caused by network latency and architectural decisions, not lack of optimization skills LLMs can fix ↓This is just superoptimization rebranded; LLMs still produce slow code and don't fundamentally change things ↓Efficient code is about memory/cache use and architecture, not something LLMs meaningfully address ~Software slowness stems from business incentives and co-tenancy constraints, not solvable by better optimization tooling