Note I believe this one because of the amount of elbow grease that went into it: 250 hours! Based on smaller projects I’ve done I’d say this post is a good model for what a significant AI-assisted systems programming project looks like.
This essay perfectly encapsulates my own experience. My biggest frustration is that the AI is astonishingly good at making awful slop which somehow works. It’s got no taste, no concern for elegance, no eagerness for the satisfyingly terse. My job has shifted from code writer to quality control officer.
Nowhere is this more obvious in my current projects than with CRUD interface building. It will go nuts building these elaborate labyrinths and I’m sitting there baffled, bemused, foolishly hoping that THIS time it would recognise that a single SQL query is all that’s needed. It knows how to write complex SQL if you insist, but it never wants to.
But even with those frustrations, damn it is a lot faster than writing it all myself.
This is the thing that gets me. The code compiles. Passes tests. So you stop reading it. Why wouldn't you.
Then three weeks later you're tracing some control flow that makes no sense and nobody knows why it's structured that way. Not you, not the model. I've been treating it like code from a contractor now, review every line same as a junior dev's PR. Gets tedious but the alternative is worse.
I’ve been treating it like a glorified autocomplete, or a glorified search and replace. Everything else is saxophone jazz when I’m writing for a string quartet: useful for inspiration, useful for understanding what isn’t clearly explained, sometimes it builds a decent first attempt, occasionally it gets shockingly close, but I’ve learned to never let my guard down. Go too far and untangling its slop becomes burdensome. Leave it to its own devices for more than a few rounds and it can become so unfixable it’s easier to start from scratch.
The "no taste" thing is real when AI is in generate-for-me mode. It's trying to fulfill your request, and won't evaluate it unsolicited. But if you change the relationship and let it react to what you're building instead of building it for you, aesthetic judgment shows up immediately. It'll tell you something is ugly or overengineered or missing the point. The taste was always in the model, it just can't express it while it's busy being obedient.
Refreshing to see an honest and balanced take on AI coding. This is what real AI-assisted coding looks like once you get past the initial wow factor of having the AI write code that executes and does what you asked.
This experience is familiar to every serious software engineer who has used AI code gen and then reviewed the output:
> But when I reviewed the codebase in detail in late January, the downside was obvious: the codebase was complete spaghetti14. I didn’t understand large parts of the Python source extraction pipeline, functions were scattered in random files without a clear shape, and a few files had grown to several thousand lines. It was extremely fragile; it solved the immediate problem but it was never going to cope with my larger vision,
Some people never get to the part where they review the code. They go straight to their LinkedIn or blog and start writing (or having ChatGPT write) posts about how manual coding is dead and they’re done writing code by hand forever.
Some people review the code and declare it unusable garbage, then also go to their social media and post how AI coding is completely useless and they’re not going to use it for anything.
This blog post shows the journey that anyone not in one of those two vocal minorities is going through right now: A realization that AI coding tools can be a large accelerator but you need to learn how to use them correctly in your workflow and you need to remain involved in the code. It’s not as clickbaity as the extreme takes that get posted all the time. It’s a little disappointing to read the part where they said hard work was still required. It is a realistic and balanced take on the state of AI coding, though.
Those extreme takes are taken mostly for clicks or are exaggerated second hand so the "other side's" opinion is dumber than it is to "slam the naysayers". Most people are meh about everything, not on the extremes, so to pander to them you mock the extremes and make them seem more likely. It's just online populism.
Agree. This is such a good balanced article. The only things that still make the insights difficult to apply to professional software development are: this was greenfield work and it was a solo project. But that’s hardly the author’s fault. It would however be fantastic to see more articles like this about how to go all in on AI tools for brownfield projects involving more than one person.
One thing I will add: I actually don’t think it’s wrong to start out building a vibe coded spaghetti mess for a project like this… provided you see it as a prototype you’re going to learn from and then throw away. A throwaway prototype is immensely useful because it helps you figure out what you want to build in the first place, before you step down a level and focus on closely guiding the agent to actually build it.
The author’s mistake was that he thought the horrible prototype would evolve into the real thing. Of course it could not. But I suspect that the author’s final results when he did start afresh and build with closer attention to architecture were much better because he has learned more about the requirements for what he wanted to build from that first attempt.
It's a very accurate and relatable post. I think one corollary that's important to note to the anti-AI crowd is that this project, even if somewhat spaghettified, will likely take orders of magnitude less time to perfect than it would for someone to create the whole thing from scratch without AI.
I often see criticism towards projects that are AI-driven that assumes that codebase is crystalized in time, when in fact humans can keep iterating with AI on it until it is better. We don't expect an AI-less project to be perfect in 0.1.0, so why expect that from AI? I know the answer is that the marketing and Twitter/LinkedIn slop makes those claims, but it's more useful to see past the hype and investigate how to use these tools which are invariably here to stay
I feel like recently HN has been seeing more takes like this one and at least slightly less of the extremist clickbaity stuff. Maybe it's a sign of maturity. (Or maybe it's just fatigue with the cycle of hyping the absolute-latest model?)
I’ve been driving Claude as my primary coding interface the last three months at my job. Other than a different domain, I feel like I could have written this exact article.
The project I’m on started as a vibe-coded prototype that quickly got promoted to a production service we sell.
I’ve had to build the mental model after the fact, while refactoring and ripping out large chunks of nonsense or dead code.
But the product wouldn’t exist without that quick and dirty prototype, and I can use Claude as a goddamned chainsaw to clean up.
On Friday, I finally added a type checker pre-commit hook and fixed the 90 existing errors (properly, no type ignores) in ~2 hours. I tried full-agentic first, and it failed miserably, then I went through error by error with Claude, we tightened up some exiting types, fixed some clunky abstractions, and got a nice, clean result.
AI-assisted coding is amazing, but IMO for production code there’s no substitute for human review and guidance.
> Some people never get to the part where they review the code. They go straight to their LinkedIn or blog and start writing (or having ChatGPT write) posts about how manual coding is dead and they’re done writing code by hand forever.
Some people review the code and declare it unusable garbage, then also go to their social media and post how AI coding is completely useless and they’re not going to use it for anything. This blog post shows the journey that anyone not in one of those two vocal minorities is going through right now.
What’s really happening is that you’re all of those people in the beginning. Those people are you as you go through the experience. You’re excited after seeing it do the impossible and in later instances you’re critical of the imperfections. It’s like the stages of grief, a sort of Kübler-Ross model for AI.
Professional software engineers like many of us have a big blind spot when it comes to AI coding, and that's a fixation on code quality.
It makes sense to focus on code quality. We're not wrong. After all, we've spent our entire careers in the code. Bad code quality slows us down and makes things slow/insecure/unreliable/etc for end users.
However, code quality is becoming less and less relevant in the age of AI coding, and to ignore that is to have our heads stuck in the sand. Just because we don't like it doesn't mean it's not true.
There are two forces contributing to this: (1) more people coding smaller apps, and (2) improvements in coding models and agentic tools.
We are increasingly moving toward a world where people who aren't sophisticated programmers are "building" their own apps with a user base of just one person. In many cases, these apps are simple and effective and come without the bloat that larger software suites have subjected users to for years. The code is simple, and even when it's not, nobody will ever have to maintain it, so it doesn't matter. Some apps will be unreliable, some will get hacked, some will be slow and inefficient, and it won't matter. This trend will continue to grow.
At the same time, technology is improving, and the AI is increasingly good at designing and architecting software. We are in the very earliest months of AI actually being somewhat competent at this. It's unlikely that it will plateau and stop improving. And even when it finally does, if such a point comes, there will still be many years of improvements in tooling, as humanity's ability to make effective use of a technology always lags far behind the invention of the technology itself.
So I'm right there with you in being annoyed by all the hype and exaggerated claims. But the "truth" about AI-assisted coding is changing every year, every quarter, every month. It's only trending in one direction. And it isn't going to stop.
I completely agree. Just going through the beginner & hobbyist forums, the change from "can you help me with code to do X" to "I used ChatGPT/Claude/Copilot to write code to do X" happened with absolutely startling speed, and it's not slowing down. There was clearly a pent-up demand here that wasn't being met otherwise.
People are using AI to get code written. They have no idea what code quality is and only care that what they built works.
AFAICT, every time technology has allowed non-technical people to do more, it's opened up new opportunities for programmers. I don't expect this to be any different, I just want to know where the opportunities are.
I'm deeply convinced that there's 2 reasons we don't see real takes like this: 1) is because these people are quietly appreciating the 2-50% uplift you get from sanely using LLMs instead of constantly posting sycophantic or doomer shit for clout and/or VC financing. 2) is because the real version of LLM coding is boring and unsexy. It either involves generating slop in one shot to POC, then restarting from scratch for the real thing or doing extensive remediation costing far more than the initial vibe effort cost; or it involves generally doing the same thing we've been doing since the assembler was created except now I don't need to remember off-hand how to rig up boilerplate for a table test harness in ${current_language}, or if I wrote a snippet with string ops and if statements and I wish it were using regexes and named capture groups, it's now easy to mostly-accurately convert it to the other form instead of just sighing and moving on.
But that's boring nerd shit and LLMs didn't change who thinks boring nerd shit is boring or cool.
It's actually common for human-written projects to go through an initial R&D phase where the first prototypes turn into spaghetti code and require a full rewrite. I haven't been through this myself with LLMs, but I wonder to what extent they could analyse the codebase, propose and then implement a better architecture based on the initial version.
For me it’s just a matter of “does this actually save me time at all?”
If it generates the slop version in a week but it takes me 3 more weeks to clean it up, could I have I just done it right the first time myself in 4 weeks instead? How much money have I wasted in tokens?
If you have something that's pretty and fast but doesn't work, is that useful?
But if you have something that works, but is ugly and slow - you can still use it and fix it while you dogfood it. Maybe figure out other ways to improve it beyond making the code look pretty.
Without wanting to sound rude: I think the mistake people make with AI prototypes is keeping the code at all.
The AI’s are more than capable of producing a mountain of docs from which to rebuild, sanely. They’re really not that capable - without a lot of human pain - of making a shit codebase good.
Someone used Claude Code to generate a very simple staffing management app. The sort of thing that really wouldn't take that long to make, but why pay for any software when you can just ignore the problem, amiright? Anyway, the code that got generated was full of SQL injection issues for the most absurd sorts of things. It would have 80% of the database queries implemented through the ORM, but then the leftover stuff was raw string concat junk, for no good reason because it wasn't even doing any dynamic query or anything that the ORM couldn't do.
Good code will stand the test of time. No matter how great a project is, there’s no world in which I adopt anything written over a few months and just released, for maintenance reasons alone.
I'm sorry, who is this for? Aren't you maybe a little tired of talking about this, if not totally <expletive> bored??
There is something at this point kind of surreal in the fact that you know everyday there will be this exact blog post and these exact comments.
Like, its been literal years and years and yall are still talking about the thing thats supposed to do other things. What are we even doing anymore? Is this dead internet? It boggles the mind we are still at this level of discourse frankly.
Love 'em hate 'em I don't care yall need to freaking get a grip! Like for the love god read a book, paint a picture! Do something else! This blog is just a journey to snooze town and we all must at some level know that. This feels like literal brain virus.
This is exactly why I built https://github.com/andonimichael/arxitect . I’ve found that agents by default produce tactical but brittle software. But if you teach agents to prioritize software architecture and design patterns, their code structure becomes much much better. Additionally, better structured code becomes more token efficient, requires less context to make changes, and coding agents become more accurate.
This blog post shows the journey that anyone not in one of those two vocal minorities is going through right now
Is there evidence these groups are a minority? I mean, the OP sounds like they are taking the right approach but I suspect it requires both skill/experience and an open mind to take their approach.
Just because an approach has good use-cases doesn't mean those are going predominate.
I find the people who end up with spaghetti code did so because they didn’t translate their normal processes over.
Being completely methodical about development really helps. obra/superpowers, for example, gets close but I think it overindexes on testing and doesn’t go far enough with design document templates, planning, code style guides, code reviews, and more.
Being methodical about it takes more time, but prevents a good bit of the tech debt.
Planning modes help, but they are similarly not methodical enough.
There is a lot you can do to shape the end result to not have these faults. In the end, the engineering mind and rigor still needs to apply, so the hard work doesn't go away.
But, the errors that are described - no architecture adhesion, lack of comprehension, random files, etc. are a matter of not leveling up the sophistication of use further, not a gap in those tools.
As an example. Very clearly laying out your architecture principles, guidance, how code should look on disk, theory on imports, etc. And then - objectively analyzing any proposed change against those principles, converges toward sane and understandable.
We've been calling it adversarial testing across a number of dimensions - architecture, security, accessibility, among other things. Every pr gets automatically reviewed and scored based on these perspectives. If an adversary doesn't OK the PR, it doesn't get merged.
i feel like it's worth noting that for a long time prior to AI, I heard a ton of anecdotes about codebases at big companies being very shitty spaghetti more often than not, and if one of the maintainers left it was often a nightmare of managing, refactoring, or in some cases re-writing whenever issues popped up with new integrations
If AI needs to re-write everything from scratch everytime you make a design change, that may have some obvious inefficiencies and limitations to it but if it can also do that in a few hours or a week, is it really that bad comapred to months of stalling and excuses from devs trying to understand the work of someone before them who wasnt given enough time to make well documented clean code to begin with?
Like it is undoubtedly worse for hobby projects to rely on the AI output 100%, but im actually not so sure for commercial products. It'll be the same type of spagetti garbage everywhere. There will be patterns in its nonsense that over years people will start to get accustomed to. Youll have people specialize in cleaning up AI generated code, and itll more or less be a relatively consistent process compared to picking up random developer spaghetti
Unfortunately there is a ton of pressure not to review code. AI enthusiasts see themselves in a race against others and code review = going slower. I recently attended a company wide "let's all use AI" conference and several talks were about how they were not reviewing much code anymore because the PRs were coming in 10x to 50x more and they just couldn't keep up.
> This blog post shows the journey that anyone not in one of those two vocal minorities is going through right now: A realization that AI coding tools can be a large accelerator but you need to learn how to use them correctly in your workflow and you need to remain involved in the code. It’s not as clickbaity as the extreme takes that get posted all the time. It’s a little disappointing to read the part where they said hard work was still required. It is a realistic and balanced take on the state of AI coding, though.
I appreciate the balanced takes and also the notion that one can use these AI tools to build software with principled use.
However, what I am still failing to see is concrete evidence that this is all faster and cheaper than just a human learning and doing everything themself or with a small team. The cat is out of the bag, so to speak, but I think it's still correct to question these things. I am putting in a _lot_ of work to reach a principled status quo with these tools, and it is still quite unclear whether it's actually improvement versus just a side quest to wrangle tools that everyone else is abusing.
Thank you. The learning aspect of reading how AI tackles something is rewarding.
It also reduces my hesitation to get started with something I don't know the answer well enough yet. Time 'wasted' on vibe-coding felt less painful than time 'wasted' on heads-down manual coding down a rabbit hole.
> architecture is what happens when all those local pieces interact, and you can’t get good global behaviour by stitching together locally correct components
This is a great article. I’ve been trying to see how layered AI use can bridge this gap but the current models do seem to be lacking in the ambiguous design phase. They are amazing at the local execution phase.
Part of me thinks this is a reflection of software engineering as a whole. Most people are bad at design. Everyone usually gets better with repetition and experience. However, as there is never a right answer just a spectrum of tradeoffs, it seems difficult for the current models to replicate that part of the human process.
I’ve had a couple wins with AI in the design phase, where it helped me reach a conclusion that would’ve taken days of exploration, if I ever got there. Both were very long conversations explicitly about design with lots of back and forth, like whiteboarding. Both involved SQL in ClickHouse, which I’m ok but not amazing at — for example I often write queries with window functions, but my mental model of GROUP BY is still incomplete.
In one of the cases, I was searching for a way to extract a bunch of code that 5-6 queries had in common. Whatever this thing was, its parameters would have to include an array/tuple of IDs, and a parameter that would alter the table being selected from, neither of which is allowed in a clickhouse parameterized view. I could write a normal view for this, but performance would’ve been atrocious given ClickHouse’s ok-but-not-great query optimizer.
I asked AI for alternatives, and to discuss the pros and cons of each. I brought up specific scenarios and asked it how it thought the code would work. I asked it to bring what it knew about SQL’s relational algebra to find the an elegant solution.
It finally suggested a template (we’re using Go) to include another sql file, where the parameter is a _named relation_. It can be a CTE or a table, but it doesn’t matter as long as it has the right columns. Aside from poor tooling that doesn’t find things like typos, it’s been a huge win, much better than the duplication. And we have lots of tests that run against the real database to catch those typos.
Maybe this kind of thing exists out there already (if it does, tell me!) but I probably wouldn’t have found it.
The 8-year wait is the part that stands out. Usually the question is "why start now" not "why did it take 8 years". Curious if there was a specific moment where the tools crossed a threshold for you, or if it was more gradual.
It's kind of click bait tho. "I took 3 months and AI to build a SQLite tool" is not going to stand out. The 8 year wait gives a sense of scale or difficulty but that's actually an illusion and does not reflect the task itself.
The author mentions a C codebase. Is AI good at coding in C now? If so, which AI systems lead in this language?
Ideally: local; offline.
Or do I have to wrestle it for 250 hours before it coughs up the dough? Last time I tried, the AI systems struggled with some of the most basic C code.
It seemed fine with Python, but then my cat can do that.
This article is describing a problem that is still two steps removed from where AI code becomes actually useful.
90 percent of the things users want either A) dont exist or B) are impossible to find, install and run without being deeply technical.
These things dont need to scale, they dont need to be well designed. They are for the most part targeted, single user, single purpose, artifacts. They are migration scripts between services, they are quick and dirty tools that make bad UI and workflows less manual and more managable.
These are the use cases I am seeing from people OUTSIDE the tech sphere adopt AI coding for. It is what "non techies" are using things like open claw for. I have people who in the past would have been told "No, I will not fix your computer" talk to me excitedly about running cron jobs.
Not everything needs to be snap on quality, the bulk of end users are going to be happy with harbor freight quality because it is better than NO tools at all.
> This article is describing a problem that is still two steps removed from where AI code becomes actually useful.
But it does a good job of countering the narrative you often see on LinkedIn, and to some extent on HN as well, where AI is portrayed as all-capable of developing enterprise software. If you spend any time in discussions hyping AI, you will have seen plenty of confident claims that traditional coding is dead and that AI will replace it soon. Posts like this is useful because it shows a more grounded reality.
> 90 percent of the things users want either A) dont exist or B) are impossible to find, install and run without being deeply technical. These things dont need to scale, they dont need to be well designed. They are for the most part targeted, single user, single purpose, artifacts.
Yes, that is a particular niche where AI can be applied effectively. But many AI proponents go much further and argue that AI is already capable of delivering complex, production-grade systems. They say, you don't need engineers anymore. They say, you only need product owners who can write down the spec. From what I have seen, that claim does not hold up and this article supports that view.
Many users may not be interested in scalability and maintainability... But for a number of us, including the OP and myself, the real question is whether AI can handle situations where scalability, maintainability and sound design DO actually matter. The OP does a good job of understanding this.
Long term, I think the best value AI gives us is a poweful tool to gain understanding. I think we are going to see deep understanding turn into the output goal of LLMs soon. For example, the blocker on this project was the dense C code with 400 rules. Work with LLMs allowed the structure and understanding to be parsed and used to create the tool, but maybe an even more useful output would be full documentation of the rules and their interactions.
This could likely be extracted much easier now from the new code, but imagine API docs or a mapping of the logical ruleset with interwoven commentary - other devtools could be built easily, bug analysis could be done on the structure of rules independent of code, optimizations could be determined on an architectural level, etc.
LLMs need humans to know what to build. If generating code becomes easy, codifying a flexible context or understanding becomes the goal that amplifies what can be generated without effort.
Looks like a clear divide in people‘s experiences based on how they use these new tools:
1) All-knowing oracle which is lightly prompted and develops whole applications from requirements specification to deployable artifacts. Superficial, little to no review of the code before running and committing.
2) An additional tool next to their already established toolset to be used inside or alongside their IDE. Each line gets read and reviewed. The tool needs to defend their choices and manual rework is common for anything from improving documentation to naming things all the way to architectural changes.
Obviously anything in between as well being viable. 1) seems like a crazy dead-end to me if you are looking to build a sustainable service or a fulfilling career.
This resonates. I had a project sitting in my head for years and finally built it in about 6 weeks recently. The AI part wasn't even the hard part honestly, it was finally commiting to actually shipping instead of overthinking the architecture. The tools just made it possible to move fast enough that I didn't lose momentum and abandon it like every other time.
This is the hardest it's ever going to be. That's been my mode for the last year. A lot of what I did in the last month was complete science fiction as little as six months ago. The scope and quality of what is possible seems to leap ahead every few weeks.
I now have several projects going in languages that I've never used. I have a side project in Rust, and two Go projects. I have a few decades experience with backend development in Java, Kotlin (last ten years) and occasionally python. And some limited experience with a few other languages. I know how to structurer backend projects, what to look for, what needs testing, etc.
A lot of people would insist you need to review everything the AI generates. And that's very sensible. Except AI now generates code faster than I can review it. Our ability to review is now the bottleneck. And when stuff kind of works (evidenced by manual and automated testing), what's the right point to just say it's good enough? There are no easy answers here. But you do need to think about what an acceptable level of due diligence is. Vibe coding is basically the equivalent of blindly throwing something at the wall and seeing what sticks. Agentic engineering is on the opposite side of the spectrum.
I actually emphasize a lot of quality attributes in my prompts. The importance of good design, high cohesiveness, low coupling, SOLID principles, etc. Just asking for potential refactoring with an eye on that usually yields a few good opportunities. And then all you need to do is say "sounds good, lets do it". I get a little kick out of doing variations on silly prompts like that. "Make it so" is my favorite. Once you have a good plan, it doesn't really matter what you type.
I also ask critical questions about edge cases, testing the non happy path, hardening, concurrency, latency, throughput, etc. If you don't, AIs kind of default to taking short cuts, only focus on the happy path, or hallucinate that it's all fine, etc. But this doesn't necessarily require detailed reviews to find out. You can make the AI review code and produce detailed lists of everything that is wrong or could be improved. If there's something to be found, it will find it if you prompt it right.
There's an art to this. But I suspect that that too is going to be less work. A lot of this stuff boils down to evolving guardrails to do things right that otherwise go wrong. What if AIs start doing these things right by default? I think this is just going to get better and better.
> Tests created a similar false comfort. Having 500+ tests felt reassuring, and AI made it easy to generate more. But neither humans nor AI are creative enough to foresee every edge case you’ll hit in the future; there are several times in the vibe-coding phase where I’d come up with a test case and realise the design of some component was completely wrong and needed to be totally reworked. This was a significant contributor to my lack of trust and the decision to scrap everything and start from scratch.
This is my experience. Tests are perhaps the most challenging part of working with AI.
What’s especially awful is any refactor of existing shit code that does not have tests to begin with, and the feature is confusing or inappropriately and unknowingly used multiple places elsewhere.
AI will write test cases that the logic works at all (fine), but the behavior esp what’s covered in an integration test is just not covered at all.
I don’t have a great answer to this yet, especially because this has been most painful to me in a React app, where I don’t know testing best practices. But I’ve been eyeing up behavior driven development paired with spec driven development (AI) as a potential answer here.
Curious if anyone has an approach or framework for generating good tests
Personally i think the challenge of testing never really changed with AI. You need to know what you want to specifically test before writing/vibe coidng with it. Otherwise it'll just manufacture tests that always passes and are of 0 value.
If some component doest benefit from being extensively tested, then it's still the same today. The difference is now it's so easy to generate something, no matter how useless it is. Worse part is, no one cares. Test passes, it doesn't affect production, line coverage increases, managers think the software is more tested, developers just let a prompt do everything. It's all just testing theatre.
I think E2E is the more important than ever. AI is pretty good at getting the local behaviour correct. So unit tests are of less value. Same can't be said for the system as a whole. The best part is, AI is actually pretty good at writing E2E tests. Ofc, given that you already know what you want to test
119 comments
[ 2.8 ms ] story [ 101 ms ] threadNowhere is this more obvious in my current projects than with CRUD interface building. It will go nuts building these elaborate labyrinths and I’m sitting there baffled, bemused, foolishly hoping that THIS time it would recognise that a single SQL query is all that’s needed. It knows how to write complex SQL if you insist, but it never wants to.
But even with those frustrations, damn it is a lot faster than writing it all myself.
Most of my questions are "in one sentence respond: long rambling context and question"
Then three weeks later you're tracing some control flow that makes no sense and nobody knows why it's structured that way. Not you, not the model. I've been treating it like code from a contractor now, review every line same as a junior dev's PR. Gets tedious but the alternative is worse.
This experience is familiar to every serious software engineer who has used AI code gen and then reviewed the output:
> But when I reviewed the codebase in detail in late January, the downside was obvious: the codebase was complete spaghetti14. I didn’t understand large parts of the Python source extraction pipeline, functions were scattered in random files without a clear shape, and a few files had grown to several thousand lines. It was extremely fragile; it solved the immediate problem but it was never going to cope with my larger vision,
Some people never get to the part where they review the code. They go straight to their LinkedIn or blog and start writing (or having ChatGPT write) posts about how manual coding is dead and they’re done writing code by hand forever.
Some people review the code and declare it unusable garbage, then also go to their social media and post how AI coding is completely useless and they’re not going to use it for anything.
This blog post shows the journey that anyone not in one of those two vocal minorities is going through right now: A realization that AI coding tools can be a large accelerator but you need to learn how to use them correctly in your workflow and you need to remain involved in the code. It’s not as clickbaity as the extreme takes that get posted all the time. It’s a little disappointing to read the part where they said hard work was still required. It is a realistic and balanced take on the state of AI coding, though.
One thing I will add: I actually don’t think it’s wrong to start out building a vibe coded spaghetti mess for a project like this… provided you see it as a prototype you’re going to learn from and then throw away. A throwaway prototype is immensely useful because it helps you figure out what you want to build in the first place, before you step down a level and focus on closely guiding the agent to actually build it.
The author’s mistake was that he thought the horrible prototype would evolve into the real thing. Of course it could not. But I suspect that the author’s final results when he did start afresh and build with closer attention to architecture were much better because he has learned more about the requirements for what he wanted to build from that first attempt.
I often see criticism towards projects that are AI-driven that assumes that codebase is crystalized in time, when in fact humans can keep iterating with AI on it until it is better. We don't expect an AI-less project to be perfect in 0.1.0, so why expect that from AI? I know the answer is that the marketing and Twitter/LinkedIn slop makes those claims, but it's more useful to see past the hype and investigate how to use these tools which are invariably here to stay
I’ve been driving Claude as my primary coding interface the last three months at my job. Other than a different domain, I feel like I could have written this exact article.
The project I’m on started as a vibe-coded prototype that quickly got promoted to a production service we sell.
I’ve had to build the mental model after the fact, while refactoring and ripping out large chunks of nonsense or dead code.
But the product wouldn’t exist without that quick and dirty prototype, and I can use Claude as a goddamned chainsaw to clean up.
On Friday, I finally added a type checker pre-commit hook and fixed the 90 existing errors (properly, no type ignores) in ~2 hours. I tried full-agentic first, and it failed miserably, then I went through error by error with Claude, we tightened up some exiting types, fixed some clunky abstractions, and got a nice, clean result.
AI-assisted coding is amazing, but IMO for production code there’s no substitute for human review and guidance.
This should be done on day one with a company-wide skill or project template that defines hard limits and processes for the Agent.
Strict linters, formatters and code quality checks are essential to de-slopify the code as much as possible.
That doesn't fix bad design though, that's still on humans.
What’s really happening is that you’re all of those people in the beginning. Those people are you as you go through the experience. You’re excited after seeing it do the impossible and in later instances you’re critical of the imperfections. It’s like the stages of grief, a sort of Kübler-Ross model for AI.
Professional software engineers like many of us have a big blind spot when it comes to AI coding, and that's a fixation on code quality.
It makes sense to focus on code quality. We're not wrong. After all, we've spent our entire careers in the code. Bad code quality slows us down and makes things slow/insecure/unreliable/etc for end users.
However, code quality is becoming less and less relevant in the age of AI coding, and to ignore that is to have our heads stuck in the sand. Just because we don't like it doesn't mean it's not true.
There are two forces contributing to this: (1) more people coding smaller apps, and (2) improvements in coding models and agentic tools.
We are increasingly moving toward a world where people who aren't sophisticated programmers are "building" their own apps with a user base of just one person. In many cases, these apps are simple and effective and come without the bloat that larger software suites have subjected users to for years. The code is simple, and even when it's not, nobody will ever have to maintain it, so it doesn't matter. Some apps will be unreliable, some will get hacked, some will be slow and inefficient, and it won't matter. This trend will continue to grow.
At the same time, technology is improving, and the AI is increasingly good at designing and architecting software. We are in the very earliest months of AI actually being somewhat competent at this. It's unlikely that it will plateau and stop improving. And even when it finally does, if such a point comes, there will still be many years of improvements in tooling, as humanity's ability to make effective use of a technology always lags far behind the invention of the technology itself.
So I'm right there with you in being annoyed by all the hype and exaggerated claims. But the "truth" about AI-assisted coding is changing every year, every quarter, every month. It's only trending in one direction. And it isn't going to stop.
I completely agree. Just going through the beginner & hobbyist forums, the change from "can you help me with code to do X" to "I used ChatGPT/Claude/Copilot to write code to do X" happened with absolutely startling speed, and it's not slowing down. There was clearly a pent-up demand here that wasn't being met otherwise.
People are using AI to get code written. They have no idea what code quality is and only care that what they built works.
AFAICT, every time technology has allowed non-technical people to do more, it's opened up new opportunities for programmers. I don't expect this to be any different, I just want to know where the opportunities are.
But that's boring nerd shit and LLMs didn't change who thinks boring nerd shit is boring or cool.
I completely agree that this is the case right now, but I do wonder how long it will remain the case.
If it generates the slop version in a week but it takes me 3 more weeks to clean it up, could I have I just done it right the first time myself in 4 weeks instead? How much money have I wasted in tokens?
2. Make it pretty
3. Make it fast.
If you have something that's pretty and fast but doesn't work, is that useful?
But if you have something that works, but is ugly and slow - you can still use it and fix it while you dogfood it. Maybe figure out other ways to improve it beyond making the code look pretty.
The AI’s are more than capable of producing a mountain of docs from which to rebuild, sanely. They’re really not that capable - without a lot of human pain - of making a shit codebase good.
There is something at this point kind of surreal in the fact that you know everyday there will be this exact blog post and these exact comments.
Like, its been literal years and years and yall are still talking about the thing thats supposed to do other things. What are we even doing anymore? Is this dead internet? It boggles the mind we are still at this level of discourse frankly.
Love 'em hate 'em I don't care yall need to freaking get a grip! Like for the love god read a book, paint a picture! Do something else! This blog is just a journey to snooze town and we all must at some level know that. This feels like literal brain virus.
Is there evidence these groups are a minority? I mean, the OP sounds like they are taking the right approach but I suspect it requires both skill/experience and an open mind to take their approach.
Just because an approach has good use-cases doesn't mean those are going predominate.
Being completely methodical about development really helps. obra/superpowers, for example, gets close but I think it overindexes on testing and doesn’t go far enough with design document templates, planning, code style guides, code reviews, and more.
Being methodical about it takes more time, but prevents a good bit of the tech debt.
Planning modes help, but they are similarly not methodical enough.
But, the errors that are described - no architecture adhesion, lack of comprehension, random files, etc. are a matter of not leveling up the sophistication of use further, not a gap in those tools.
As an example. Very clearly laying out your architecture principles, guidance, how code should look on disk, theory on imports, etc. And then - objectively analyzing any proposed change against those principles, converges toward sane and understandable.
We've been calling it adversarial testing across a number of dimensions - architecture, security, accessibility, among other things. Every pr gets automatically reviewed and scored based on these perspectives. If an adversary doesn't OK the PR, it doesn't get merged.
If AI needs to re-write everything from scratch everytime you make a design change, that may have some obvious inefficiencies and limitations to it but if it can also do that in a few hours or a week, is it really that bad comapred to months of stalling and excuses from devs trying to understand the work of someone before them who wasnt given enough time to make well documented clean code to begin with?
Like it is undoubtedly worse for hobby projects to rely on the AI output 100%, but im actually not so sure for commercial products. It'll be the same type of spagetti garbage everywhere. There will be patterns in its nonsense that over years people will start to get accustomed to. Youll have people specialize in cleaning up AI generated code, and itll more or less be a relatively consistent process compared to picking up random developer spaghetti
maybe this is a hot take though
Scary AF
I appreciate the balanced takes and also the notion that one can use these AI tools to build software with principled use.
However, what I am still failing to see is concrete evidence that this is all faster and cheaper than just a human learning and doing everything themself or with a small team. The cat is out of the bag, so to speak, but I think it's still correct to question these things. I am putting in a _lot_ of work to reach a principled status quo with these tools, and it is still quite unclear whether it's actually improvement versus just a side quest to wrangle tools that everyone else is abusing.
It also reduces my hesitation to get started with something I don't know the answer well enough yet. Time 'wasted' on vibe-coding felt less painful than time 'wasted' on heads-down manual coding down a rabbit hole.
This is a great article. I’ve been trying to see how layered AI use can bridge this gap but the current models do seem to be lacking in the ambiguous design phase. They are amazing at the local execution phase.
Part of me thinks this is a reflection of software engineering as a whole. Most people are bad at design. Everyone usually gets better with repetition and experience. However, as there is never a right answer just a spectrum of tradeoffs, it seems difficult for the current models to replicate that part of the human process.
In one of the cases, I was searching for a way to extract a bunch of code that 5-6 queries had in common. Whatever this thing was, its parameters would have to include an array/tuple of IDs, and a parameter that would alter the table being selected from, neither of which is allowed in a clickhouse parameterized view. I could write a normal view for this, but performance would’ve been atrocious given ClickHouse’s ok-but-not-great query optimizer.
I asked AI for alternatives, and to discuss the pros and cons of each. I brought up specific scenarios and asked it how it thought the code would work. I asked it to bring what it knew about SQL’s relational algebra to find the an elegant solution.
It finally suggested a template (we’re using Go) to include another sql file, where the parameter is a _named relation_. It can be a CTE or a table, but it doesn’t matter as long as it has the right columns. Aside from poor tooling that doesn’t find things like typos, it’s been a huge win, much better than the duplication. And we have lots of tests that run against the real database to catch those typos.
Maybe this kind of thing exists out there already (if it does, tell me!) but I probably wouldn’t have found it.
Ideally: local; offline.
Or do I have to wrestle it for 250 hours before it coughs up the dough? Last time I tried, the AI systems struggled with some of the most basic C code.
It seemed fine with Python, but then my cat can do that.
90 percent of the things users want either A) dont exist or B) are impossible to find, install and run without being deeply technical.
These things dont need to scale, they dont need to be well designed. They are for the most part targeted, single user, single purpose, artifacts. They are migration scripts between services, they are quick and dirty tools that make bad UI and workflows less manual and more managable.
These are the use cases I am seeing from people OUTSIDE the tech sphere adopt AI coding for. It is what "non techies" are using things like open claw for. I have people who in the past would have been told "No, I will not fix your computer" talk to me excitedly about running cron jobs.
Not everything needs to be snap on quality, the bulk of end users are going to be happy with harbor freight quality because it is better than NO tools at all.
But it does a good job of countering the narrative you often see on LinkedIn, and to some extent on HN as well, where AI is portrayed as all-capable of developing enterprise software. If you spend any time in discussions hyping AI, you will have seen plenty of confident claims that traditional coding is dead and that AI will replace it soon. Posts like this is useful because it shows a more grounded reality.
> 90 percent of the things users want either A) dont exist or B) are impossible to find, install and run without being deeply technical. These things dont need to scale, they dont need to be well designed. They are for the most part targeted, single user, single purpose, artifacts.
Yes, that is a particular niche where AI can be applied effectively. But many AI proponents go much further and argue that AI is already capable of delivering complex, production-grade systems. They say, you don't need engineers anymore. They say, you only need product owners who can write down the spec. From what I have seen, that claim does not hold up and this article supports that view.
Many users may not be interested in scalability and maintainability... But for a number of us, including the OP and myself, the real question is whether AI can handle situations where scalability, maintainability and sound design DO actually matter. The OP does a good job of understanding this.
This could likely be extracted much easier now from the new code, but imagine API docs or a mapping of the logical ruleset with interwoven commentary - other devtools could be built easily, bug analysis could be done on the structure of rules independent of code, optimizations could be determined on an architectural level, etc.
LLMs need humans to know what to build. If generating code becomes easy, codifying a flexible context or understanding becomes the goal that amplifies what can be generated without effort.
1) All-knowing oracle which is lightly prompted and develops whole applications from requirements specification to deployable artifacts. Superficial, little to no review of the code before running and committing.
2) An additional tool next to their already established toolset to be used inside or alongside their IDE. Each line gets read and reviewed. The tool needs to defend their choices and manual rework is common for anything from improving documentation to naming things all the way to architectural changes.
Obviously anything in between as well being viable. 1) seems like a crazy dead-end to me if you are looking to build a sustainable service or a fulfilling career.
Expanding a thought beyond 280 characters and publishing it somewhere other than the X outrage machine is something we should be encouraging.
Seconded!
I now have several projects going in languages that I've never used. I have a side project in Rust, and two Go projects. I have a few decades experience with backend development in Java, Kotlin (last ten years) and occasionally python. And some limited experience with a few other languages. I know how to structurer backend projects, what to look for, what needs testing, etc.
A lot of people would insist you need to review everything the AI generates. And that's very sensible. Except AI now generates code faster than I can review it. Our ability to review is now the bottleneck. And when stuff kind of works (evidenced by manual and automated testing), what's the right point to just say it's good enough? There are no easy answers here. But you do need to think about what an acceptable level of due diligence is. Vibe coding is basically the equivalent of blindly throwing something at the wall and seeing what sticks. Agentic engineering is on the opposite side of the spectrum.
I actually emphasize a lot of quality attributes in my prompts. The importance of good design, high cohesiveness, low coupling, SOLID principles, etc. Just asking for potential refactoring with an eye on that usually yields a few good opportunities. And then all you need to do is say "sounds good, lets do it". I get a little kick out of doing variations on silly prompts like that. "Make it so" is my favorite. Once you have a good plan, it doesn't really matter what you type.
I also ask critical questions about edge cases, testing the non happy path, hardening, concurrency, latency, throughput, etc. If you don't, AIs kind of default to taking short cuts, only focus on the happy path, or hallucinate that it's all fine, etc. But this doesn't necessarily require detailed reviews to find out. You can make the AI review code and produce detailed lists of everything that is wrong or could be improved. If there's something to be found, it will find it if you prompt it right.
There's an art to this. But I suspect that that too is going to be less work. A lot of this stuff boils down to evolving guardrails to do things right that otherwise go wrong. What if AIs start doing these things right by default? I think this is just going to get better and better.
This is my experience. Tests are perhaps the most challenging part of working with AI.
What’s especially awful is any refactor of existing shit code that does not have tests to begin with, and the feature is confusing or inappropriately and unknowingly used multiple places elsewhere.
AI will write test cases that the logic works at all (fine), but the behavior esp what’s covered in an integration test is just not covered at all.
I don’t have a great answer to this yet, especially because this has been most painful to me in a React app, where I don’t know testing best practices. But I’ve been eyeing up behavior driven development paired with spec driven development (AI) as a potential answer here.
Curious if anyone has an approach or framework for generating good tests
If some component doest benefit from being extensively tested, then it's still the same today. The difference is now it's so easy to generate something, no matter how useless it is. Worse part is, no one cares. Test passes, it doesn't affect production, line coverage increases, managers think the software is more tested, developers just let a prompt do everything. It's all just testing theatre.
I think E2E is the more important than ever. AI is pretty good at getting the local behaviour correct. So unit tests are of less value. Same can't be said for the system as a whole. The best part is, AI is actually pretty good at writing E2E tests. Ofc, given that you already know what you want to test