Hacker News new | ask | show | jobs
by claytongulick 105 days ago
Serious question here.

Have you taken a moment to step back and truly evaluate your productivity when using LLMs for code generation?

I don't mean the obvious confirmation-bias tickling stuff like "create a form with these fields and validation".

I mean from a whole-system, total effort analysis, from idea to production, support and maintenance.

I'm curious what you find.

My current theory is that the industry will land in a place where LLMs for code generation are frowned upon for non-trivial work, but that they are embraced for tooling, summarization and explanation.

I think these things have real, concrete value, but that it's a mistake to substitute them for human reasoning - and human reasoning is a crucial characteristic of quality code.

The thing I'm not sure of is whether the current "good enough is good enough" approach to vibe coded solutions will be sticky, or in what contexts.

MS Access still powers entire business departments, because good enough is good enough.

2 comments

This resonates. The problem I keep running into isn't that the model is bad — it's that the feedback loop is too thin. A y/n in the terminal isn't enough to catch when the model does something subtly wrong.
I've been building a review UI layer for coding agents (Claude Code, Codex) that lets you actually inspect and edit what the agent is about to do before it executes: https://github.com/agentlayer-io/AgentClick

Turns out most of the "dumb" mistakes OP is talking about are catchable — you just need to actually see them before they ship.

For frontend code it's fine tbh. Perfectly capable with oversight