Live data from Hacker News

Ask HN: How can I get better at using AI for programming?

news.ycombinator.com

451–460 of 491 posts

Re: Ask HN: How can I get better at using AI for programming?

#451

Earlier quoted context omitted.

No the point was quite clear: > A few years ago we didn't have an imprecise nondeterministic programming language that would allow your mom to achieve SOTA results on a wide range of NLP tasks by asking nicely, or I'm sure people would have taken it. But that (accurate) point makes your point invalid, so you'd rather focus on the dressing.

We still don't have that programming language (although "SOTA" and "wide range of NLP tasks" are vague enough that you can probably move the goalposts into field goal range).

This comment is written way too adversarially for someone who doesn't know what NLP is.

Re: Ask HN: How can I get better at using AI for programming?

#454

Hey, I am the co-founder at https://brew.studio We built the platform for the problem you are facing. Our agent `brew` excels at managing context with growing change requests and actual code change so that solving spaghetti code is easier and efficient. We will allow you to create and supervise impact analysis on a change request you are planning and then you get brew to also create a detailed implementation prompt f…

Brew? Didn't anyone tell you that name was taken already?

Re: Ask HN: How can I get better at using AI for programming?

#455
post #421

Earlier quoted context omitted.

Whether its blast furnaces or carbon fiber, the wear and tear (macroscopic changes) as well as material fatigue (molecular changes) is something that will be specified by the manufacturer, within some margin of error and you pretty much know what to expect - unless you are a smartass billionaire building an improvised sub off of carbon fiber whose expiry date was long due. However, the carbon fiber or your blast furn…

Why would I agree that "LLMs are shit for any serious work" when I've been using them for serious work for two+ years, as have many other people who's skills I respected from before LLMs came along? I wrote about another solid case study this morning: https://simonwillison.net/2025/Dec/14/justhtml/ I genuinely don't understand how you can look at all of this evidence and still conclude that they aren't useful for peo…

Well, you dont have to agree with that statement. But I havent seen a serious refute of my arguments either.

Re: Ask HN: How can I get better at using AI for programming?

#456

Earlier quoted context omitted.

I don't think that long, but yeah, I give it five years. Two years and 3/4 will be not needed anymore

I don't have all the variables in (financials of openai debt etc) but a few articles mention that they leverage part of their work to {claude,gemini,chatgpt} code agents internally with good results. it's a first step in a singularity like ramp up. People think they'll have jobs maintaining AI output but i don't see how maintaining is that harder than creating for a llm able to digest requirements and codebase and it…

I don't think either, people forget that agents are also developing.

Back then, we put all the source code into AI to create things, then we manually put files into context, now it looks for needed files on their own. I think we can do even better by letting AI create a file and API documentation and only read the file when really needed. And select the API and documentation it needs and I bet there is more possible, including skills and MCP on top.

So, not only LLMs are getting better, but also the software using it.

Re: Ask HN: How can I get better at using AI for programming?

#457
post #42
post #15

Earlier quoted context omitted.

That's a fun idea. How do you get the transcript into Claude Code (or whatever you use)? What transcription service do you use?

There are a few apps nowadays for voice transcription. I've used Wispr Flow and Superwhisper, and both seem good. You can map some hotkey (e.g., ctrl + windows) to start recording, then when you press it again to stop, it'll get pasted into whatever text box you have open Superwhisper offers some AI post-processing of the text (e.g., making nice bullets or grammar), but this doesn't seem necessary and just makes thin…

+1 for Superwhisper. It has an offline model for transcription. And it transcribes with very high accuracy for me and great speed.

Re: Ask HN: How can I get better at using AI for programming?

#458

Here’s how I would do this task with cursor, especially if there are more routes. I would open a chat and refactor the template together with cursor: I would tell it what I want and if I don’t like something, I would help it to understand what I like and why. Do this for one route and when you are ready, ask cursor to write a rules file based on the current chat that includes the examples that you wanted to change an…

Yes, I do things like this too. Planning works great, but the same is true for making write-ups after completing a task. In this case you'd ask the agent to write up a document with all the rules for your refactoring based on the current thread (where you have discovered the rules with the agent as you go). Let it include examples and let it pause and report back to you if it comes across a situation that isn't fully covered by the current rules yet so you can review it ,update the rules, and then let it continue

Re: Ask HN: How can I get better at using AI for programming?

#459
post #15

Earlier quoted context omitted.

That's a fun idea. How do you get the transcript into Claude Code (or whatever you use)? What transcription service do you use?

made this tool to press double control to start and another ctrl to stop which copies to the cliboard https://github.com/elv1n/para-speak/

So cool man! Had to add couple fixes to be able to use it on mac. Love it!

Re: Ask HN: How can I get better at using AI for programming?

#460
I am a day late to this conversation: while I sometimes find full-on agent-first based coding assistants like Claude Code, Google Antigravity, OpenAI codex, etc. useful, I think it is a mistake giving away too much of our autonomy as developers to default to agent-first.

I like to start any new code, or modification task by writing out design and possible implementation notes. I have a few simple command line tools that will combine one of my hand crafted notes files with code into one big beautiful prompt, submit it as a one-shot request. Then I review and use the results, doing an iteration if required.

Absolutely there are times when I feel that first dropping into something like Antigravity or Claude Code is the best approach, but I don’t want that as my default.

Post reply on HN