I created my first AI-assisted pull request
nelsonfigueroa
67 points
69 comments
March 24, 2026
Related Discussions
Found 5 related stories in 45.6ms across 3,471 title embeddings via pgvector HNSW
- A standard protocol to handle and discard low-effort, AI-Generated pull requests Muhammad523 · 138 pts · March 05, 2026 · 57% similar
- Things I've Done with AI shepherdjerred · 80 pts · March 09, 2026 · 55% similar
- 6 Practices that turned AI from prototyper to workhorse (106 PRs in 14 days) waleedk · 15 pts · March 01, 2026 · 49% similar
- I built an AI receptionist for a mechanic shop mooreds · 245 pts · March 23, 2026 · 49% similar
- I used AI. It worked. I hated it signa11 · 11 pts · April 03, 2026 · 48% similar
Discussion Highlights (18 comments)
winrid
Yeah I mean, now you know how managers feel? :) spend all day talking to people (except it's LLMs) and not sure if you accomplished anything, but people seem happy The plus side is for your personal things like this you don't have to use it of course!
tptacek
At the end of the day, the shareholders care about delivering features, gaining customers, and making money. They don’t care how software is built. They absolutely do care how software is built. They just don't weight the factors the same way you do. Product companies exist to convert software into money by providing utility to users. There's really no part of the transaction that meaningfully involves how much fun you're personally having building it.
ilc
When you realize that being a great "programmer", isn't about writing the most code, but getting the job done... AI will click as another tool in the toolbox.
largbae
This is the wrong way to think about tool use. You wanted this feature for years. You understood the problem, but the amount of time that it would have taken to properly implement and test it held you back from doing it. Obviously, anyone else who wanted this feature came to the same conclusion. This new tool reduced the amount of time that it would take. So you used the tool. You used the tool to bring the feature into existence, checked the tests, and took enough time to ensure that it was good. You didn't lie about your contribution in the PR, and the maintainer deemed it acceptable. And now everyone has this feature! When you eat a strawberry do you feel like an impostor for not growing it yourself?
flashgordon
Wait till you hit your 1000th. And its not just "assisted" but lock stock and barelled!
Spooky23
My perspective on this is that my first real job in 1999 was a DBA. I was an intern and then junior focused on the Oracle and Informix database and optimization of the systems and storage. Basically the Unix sysadmin who grokked database. We had 8 people on that team. The entire scope of what we did for a living was replaced, mostly by 2010 or so. My role was made redundant by improving storage performance and capacity. We had a few TB and lots of blob data. I cared about where data was stored from a disk geometry perspective. Today, I could smoke that infrastructure with my MacBook. The other DBA roles also mostly moved on. ORMs automated a lot of schema work. Engine optimizations eliminated a lot of the operational tuning work that went on. Most of the other stuff moved into adjacent developer roles. Most places have very few DBAs today. That startup today would have had zero. I think the author is being way too hard on himself. He defined a problem, worked with the computer to “scratch the itch” presumably QA’d the result and sent it upstream. That’s valid and useful. The method is different. But the work is solving the problems - and just like crazy kids solved problems with VisualBasic and the real men wielding C++ shook their heads, the AI tools are going to produce alot of shit, but also solve alot of problems.
zem
my advice for anyone in this situation - as the next step. work with claude to understand the context surrounding the change, and how it fits into the existing project. do a deeper dive into whether the change causes a performance regression, see if it's stylistically consistent with similar code, probe whether there is code duplication that you could clean up by pulling out a function. claude can absolutely help with all of this! and at the end of all that, read through the generated code with a fresh understanding of it, and see if you agree with how it was done or whether you might prefer anything to be changed. you will get a proper sense of ownership and of at least having put some work into not delivering slop, though of course there might still be subtle issues that only the people familiar with the codebase would catch.
prakashrj
I am learning more than before. It helps me build things faster. You can still make it robust, efficient, crafty, etc. You have to improve your AI-assisted coding knowledge to achieve all those things.
acedTrex
Ya its a shitty feeling
whatever1
LLMs have changed completely the time economics of coding. Things that in the past you would never touch because they would not worth the time investment of yours, today can be almost be done in one shot by LLMs. I built a TV OS slideshow app for both photos and videos (as far as I know all the apps just go through photos). I have no experience in Apple OSes development and in the past it would had taken me at least a week to just read enough documentation to get started. Now? It took me 3 hours of iterating with an LLM to start from scratch developing and publish the app.
tayo42
I feel like I need to build a rocket ship or something something to feel like I programmed something cool and feel good about it.
kjgkjhfkjf
It seems like the author feels like a fraud because they successfully made a contribution without learning anything. The quality of a contribution is not a function of how much you learned or grew while you made it. Learning and growth are part of your compensation for making the contribution. The author is not a fraud for not learning anything. If anything, it seems like they should feel short-changed! When voluntarily making contributions to open-source projects, everyone should of course feel free not to use AI tools if they want. However, I would argue that using AI tools is a valuable skill itself, and worth practicing.
6thbit
> I didn’t learn anything. I felt like I was flinging slop over the wall to an open-source maintainer. Well I’m sorry you feel that way, impostor syndrome is tough to deal with already without AI. You seem to be driven by understanding and you have a great tool to learn from here if you make an effort over time to grasp the “slop” you’re throwing to the wall. Be curious, ask why several times and explore guilt free over time when you are in the right mindset. I’m glad you got something useful out of it this time and also not everything you do with AI has to be useful or a final “deliverable”, it can also be a great toy and window into more understanding.
skeptrune
Part of why I originally picked programming as a profession is because it felt like a way to get paid to be more of an artisan craftsman. Ultimately, I don't think it was ever quite that, even before ai. But now post LLM coding agents, its not at all that. Nothing about programming for money resembles artisanship. It might be time to try sewing wallets or something...
mayukh
I just typed this in and hit send. I feel like a fraud.
automatoney
I feel similarly to the author, and I appreciate the links to other authors with similar sentiments - "hirirng a taskrabbit to solve a puzzle" or "feeling nothing about the results". I don't enjoy using LLMs, and I'm sad that it feels like we are a shrinking minority of programmers. It feels like a growing gulf that I increasingly don't want to try and reach across - I do not have fun with this tool, and I don't really want to hang around somewhere people are frequently pedaling it. At the very least the change has made me reduce the amount of time I spend here. But I'm still a bit bummed about it.
sublinear
When I read stuff like this, I get concerned about how big the knowledge gaps for some people really are. How crazy is the time pressure that reviewing the code is a big deal? What is there to "learn"? That's not what code review is about. As long as you understand it before committing, you own your version of it now. There's no way in hell I'd waste time playing the slot machine. I am perfectly capable of writing the exact missing parts I need to integrate and move on quickly. How is this any different from SO copypasta a decade ago? Just like that wasn't always the right tool for the job, neither is "AI". This sounds like a completely different problem than AI usage itself. My time is most valuable making decisions for the project. Yes, the vast majority of those decisions involve the code for the implementation details, but I just need clean simple code that does the job and that anyone or anything can easily change later. The AI doesn't always give that to me, and sometimes neither do other humans. That's why I'm employed. That's what it really means to be a maintainer and contribute.
annjose
I love the description of the PR. This type of honest statement is the right thing to do - be transparent, be respectful of the time of the reviewer. > This PR adds support for embedded Ruby (ERB) which is commonly used in Ruby on Rails projects. Note that I used heavy assistance from Claude Code and tried to ensure it didn't generate slop to the best of my abilities. All tests are passing and I also visually verified the end result which looks solid to me. > Here's a screenshot that was generated by building the Chroma CLI with the ERB lexer and running it against the test data file with chroma --lexer=erb --style=monokai --html lexers/testdata/erb.actual