[rco8786]

I feel like I do all of this stuff and still end up with unusable code in most cases, and the cases where I don't I still usually have to hand massage it into something usable. Sometimes it gets it right and it's really cool when it does, but anecdotally for me it doesn't seem to be making me any more efficient.

[enobrev]

> it doesn't seem to be making me any more efficient

That's been my experience.

I've been working on a 100% vibe-coded app for a few weeks. API, React-Native frontend, marketing website, CMS, CI/CD - all of it without changing a single line of code myself. Overall, the resulting codebase has been better than I expected before I started. But I would have accomplished everything it has (except for the detailed specs, detailed commit log, and thousands of tests), in about 1/3 of the time.

[fourthark]

How long would it have taken if you had written “the detailed specs, detailed commit log, and thousands of tests”?

[veber-alex]

-1 time because it would have never have happened without AI

[enobrev]

The specs would not likely have happened at all, since this is a solo project; although this experience has led me to want to write these things out more thoroughly, even for myself. It's impressive how little work I need to put in going this route to have fairly thorough actionable specs for pretty much every major decision I've made through the process.

The commits - some would be detailed, plenty would have been "typo" or "same as last commit, but works this time"

The tests - Probably would have been decent for the API, but not as thorough. Likely non-existent for the UI.

As for time - I agree with the other response - I wouldn't have taken the time.

[theshrike79]

I'm not calling bullshit here, but something smells.

If you really can write a full-ass system like that faster than an LLM, you're either REALLY fucking good at what you do (and an amazing typer), or you're holding the LLM wrong as they say.

[enobrev]

I'm ok on speed. Not 10x or anything, but I've been writing full-stack web apps and websites from scratch for a quarter century.

The issue is getting the LLM to write _reasonably decent_ code without having to read every line and make sure it's not doing anything insane. I've tried a few different methods of prompting, but setting up a claude sub-agent that's doing TDD very explicitly and ensuring that all tests pass after every iteration has been most effective.

My first attempt was so fast, it was mind-bending. I had a "working" App and API running in about a day and a half. And then when I tried to adjust features, it would start changing things all over the place, LOTS of tests were failing, and after a couple prompts, I got to a point where the app was terribly broken. I spent about a day trying to prompt my way out of a seemingly infinite hole. I did a more thorough code review and it was a disaster: Random code styles, tons of half-written and abandoned code, tests that did nothing, //TODOs everywhere, and so, so many tweaks for backwards compatibility - which I did NOT need for a greenfield project

At that point I scrapped the project and adjusted my approach. I broke down the PRD into more thorough documentation for reference. I streamlined the CLAUDE.md files. I compiled a standard method of planning / documenting work to be done. I created sub-agents for planning and implementation. I set up the primary implementation sub-agents to split up the spec into bite-sized portions of work ("30-45 minute tasks").

Now I'm at the opposite side of the spectrum - implementation is dog slow, but I rarely have to read what was actually written. I still review the code at large after the primary tasks are finished (comparing the feature branch against main in my IDE), but for the most part I've been able to ignore the output and rely on my manual app tests and then occasionally switch models (or LLMs) and prompt for a thorough code-review.

[jaggederest]

The key is prompting. Prompt to within an inch of your life. Treat prompts as source code - edit them in files, use @ notation to bring them into the console. Use Claude to generate its own prompts - https://github.com/wshobson/commands/ and https://github.com/wshobson/agents/ are very handy, they include a prompt-engineer persona.

I'm at the point now where I have to yell at the AI once in a while, but I touch essentially zero code manually, and it's acceptable quality. Once I stopped and tried to fully refactor a commit that CC had created, but I was only able to make marginal improvements in return for an enormous time commitment. If I had spent that time improving my prompts and running refactoring/cleanup passes in CC, I suspect I would have come out ahead. So I'm deliberately trying not to do that.

I expect at some point on a Friday (last Friday was close) I will get frustrated and go build things manually. But for now it's a cognitive and effort reduction for similar quality. It helps to use the most standard libraries and languages possible, and great tests are a must.

Edit: Also, use the "thinking" commands. think / think hard / think harder / ultrathink are your best friend when attempting complicated changes (of course, if you're attempting complicated changes, don't.)

[thayne]

This works fairly well for well defined, repetitive tasks. But at least for me, if you have to put that much effort into the prompt, it is likely easier just to write the code myself.

[masto]

Sometimes I spend half an hour writing a prompt and realize that I’ve basically rubber-ducked the problem to the point where I know exactly what I want, so I just write the code myself.

I have been doing my best to give these tools a fair shake, because I want to have an informed opinion (and certainly some fear of being left behind). I find that their utility in a given area is inversely proportional to my skill level. I have rewritten or fixed most of the backend business logic that AI spits out. Even if it’s mostly ok on a first pass, I’ve been doing this gig for decades now and I am pretty good at spotting future technical debt.

On the other hand, I’m consistently impressed by its ability to save me time with UI code. Or maybe it’s not that it saves me time, but it gets me to do more ambitious things. I’d typically just throw stuff on the page with the excuse that I’m not a designer, and hope that eventually I can bring in someone else to make it look better. Now I can tell the robot I want to have drag and drop here and autocomplete there, and a share to flooberflop button, and it’ll do enough of the implementation that even if I have to fix it up, I’m not as intimidated to start.

[theshrike79]

I've had the Corporate Approved CoPilot + Sonnet 4 write a full working React page for me based on a screenshot of a Figma model. (Not even through an MCP)

It even discovered that we have some internal components and used them for it.

Got me from 0-MVP in less then an hour. Would've easily taken me a full day

[NitpickLawyer]

I've found it works really well for exploration as well. I'll give it a new library, and ask it to explore the library with "x goal" in mind. It then goes and agents away for a few minutes, and I get a mini-poc that more often than not does what I wanted and can also give me options.

[xenobeb]

I am certain it has much to do with being in the training data or not.

I have loved GPT5 but the other day I was trying to implement a rather novel idea that would be a rather small function and GPT5 goes from a genius to an idiot.

I think HN has devolved into random conversations based on a random % of problems being in the training data or not. People really are having such different experiences with the models based on the novelty of the problems that are being solved.

At this point it is getting boring to read.

[rco8786]

Have you made any attempt to quantify your efficiency/output vs writing the code yourself? I've done all of these things you've mentioned, with varying degrees of success. But also everything you're talking about doing is time consuming and eats away at whatever efficiency gain CC claims to offer.

[jaggederest]

Days instead of weeks, basically. Hard to truly quantify but I'm bloody minded enough to reimplement things three times to check and even with foresight the AI is faster.

[shaunxcode]

I am convinced that this comment once read aloud in the cadence of Ginsberg is a work of art!

[jaggederest]

Now I'm trying to find a text-to-Ginsberg translator. Maybe he's who I sound like in my head.

[fragmede]

How much voice control have you implemented?

[jaggederest]

None but it's on the list! Actually using it to prototype a complete audio visual tracking and annotation tool, so feeding it back into itself is a logical next step.