3uler

AI is pretty good at following existing patterns in a codebase. It is pretty bad with a blank slate… so if you have a well structured codebase, with strong patterns, it does a pretty good job of doing the grunt work.

I mean sure… but to me that is as likely as the official ui misrepresenting the info.

The read-only hesitation seems overcautious. If you’re genuinely using it read-only, what’s the failure mode? The tool crashes or returns bad data - same risks as the AWS CLI or console.

The “middleware layer” concern doesn’t hold up. This is just a better interface for exploring AWS resources, same as k9s is for Kubernetes. If you trust k9s (which clearly works, given how widely it’s used), the same logic applies here.

If you’re enforcing infrastructure changes through IaC, having a visual way to explore your AWS resources makes sense. The AWS console is clunky for this.

The fakes vs mocks distinction here feels like a terminology debate masking violent agreement. What you’re describing as a “fake” is just a well-designed mock. The problem isn’t mocks as a concept, it’s mocking at the wrong layer. The rule: mock what you own, at the boundaries you control. The chaos you describe comes from mocking infrastructure directly. Verifying “deleteUserById was called exactly once with these params” is testing implementation, not behavior. Your HashMap-backed fake tests the right thing: is the user gone after the operation? Who cares how. The issue is finding the correct layers to validate behavior, not the implementation detail of mocks or fakes… that’s like complaining a hammer smashed a hole in the wall.

Google had a good 10 year run, where the ads were genuinely useful, until the need of the public markets required and lack of competition allowed them to enshitify the experience to the current state.

I hope the same fate does not await ChatGPT but in the mean time I expect it to be a pretty good experience at first.

Comments it seems taken in bad faith

I mean the sort of user you are describing sounds like they’d struggle with PC gaming in general.

I've always maintained that microservices are more for scaling people than scaling systems.

But then they’ve not reviewed it themselves?

The lady doth protest too much. People see every AI limitation crystal clear, but zero self awareness of their own fallibility.

and that is what makes it a great language for LLM's to generate, just don't make me write it...

Yes it is a skill issue - I lack the skill to enjoy programming in a language with the ergonomics of something from the 70s. Golang is C with garbage collection

They’re complaining about mediocre AI-generated Go code, when Go was explicitly designed to optimize for mediocrity at scale. Rob Pike literally said they designed it for programmers who “are not capable of understanding a brilliant language.” The language deliberately trades expressiveness for simplicity so that huge teams of junior engineers can’t shoot themselves in the foot.

LLMs are basically junior engineers with perfect syntax recall. Of course they generate Go well, verbose, explicit, no clever abstractions. That’s not a bug, it’s the entire design philosophy.

For most of my work, TypeScript/Node is plenty fast and I can work fullstack in one language. When I actually need performance, Rust gives me control without random GC pauses. And if I need a GC language with good ergonomics, Kotlin on the JVM is miles ahead.

Go made sense in 2010 when Google needed to get thousands of new grads productive quickly. But those tradeoffs, sacrificing language quality for organizational scale - are exactly why it’s perfect for AI generation. You can’t have it both ways: you can’t design a language for the lowest common denominator and then be surprised when AI hits exactly that bar.

I hate golang as language I just cannot get over how much I hate its syntax and I hate how verbose it is… however I do love that it is fast, compiles to a single binary and has a pretty nice standard library.

LLMs are the only way for me to make go usable.

The idea of “nice”, “high-quality” golang is an oxymoron. The very nature of the language makes it impossible to write nice high quality code… it’s designed by big tech to get college grads to pump out reams of garbage as fast as possible. LLMs are about as smart as college grads, so It was literally designed for LLMs to generate!

The Electoral Roll is quite different though

https://en.wikipedia.org/wiki/Electoral_roll

When I moved to the Netherlands I was shocked to find out you have to maintain a registered address with the government.

The government also decides how many non-family members can register at an address, so in Amsterdam it is common for people to remain registered at there parents while subletting a room in an apartment.

You also get a DigiD which very convenient but also terrifying, especially when I walk around my neighborhood and see plaque’s in the ground for the victims of the holocaust who lived here.

My Dutch girlfriend does not believe me when I tell here that you don’t have to register where you live with the government in the anglophone world. It’s just so engrained in the society that anything else seems absurd.

It’s less about the models getting smarter and more about them getting better at handling vague requests and context acquisition. They’re better at figuring out what they need to know, I’m better at shaping that process, and I have structured workflows for managing and efficiently feeding the right context into each prompt.

The problem is that there is a lot of bad python and typescript/javascript out there, and I similarly find my self having to define my coding style in context files in order to get decent results in newer code bases without a lot of examples to follow. And even then you need to say do it like @example.py all the time.

Maybe the future is fine-tuned models on specific coding styles?

That’s like saying a vim expert would be slower in VS Code - technically the IDE does more, but expertise with your existing tools often beats learning new ones.

Also that study was from early 2025 before Claude 4 which to me was a big break through in productivity, I did not really find these tools too useful before using sonnet 4.

Many of the frameworks mentioned leverage subagents

I never found taskmaster that useful, something about how it forced you to work didn’t click with me…

It’s basically a set of commands and agents and a way to structure context.

The BMAD system seems similar to the AgentOS mentioned in the post.

This way of context engineering has definitely been the way to go for me, although I’ve just implemented it myself… using Claude to help generate commands and agents and tweaking them to my liking, lately been using json as well as markdown to share context between steps.

The UV index in the southern hemisphere goes a lot higher than anything you experience up in the northern hemisphere. Do yourself a favour and go have look at the UV index on a hot summers day in Sydney in January.

So does trello, they are not good anymore.

Atlasian are the masters of inshitification/jirafication of the products they buy. Turning loved products into hated gartner matrixes.

Another excellent product off to the gallows at atlasian. Joining ops genie and trello.

This is exactly how you bootstrap payment rails to compete with Visa/MC. Merchants would do anything to reduce the 2-3% they’re bleeding to the card schemes. Everyone focuses on consumer adoption, but merchants push payment methods customers don’t love all the time - ACH transfers, store cards, cash discounts. If stablecoins can cut interchange from 2-3% to near zero, merchants will drive adoption through discounts and incentives. Gas stations where card fees destroy margins, high-volume retailers - get a few major players offering meaningful stablecoin discounts and suddenly consumers have a reason to figure out the wallets.

Yes but bitcoin is essentially useless as an unit of exchange because it’s extremely unstable and deflationary nature. The only “logical” thing to do with it is to HODL.

Working with LLMs has fundamentally changed how I approach documentation and development.

Traditional documentation has always been a challenge for me - figuring out where to start, what syntax conventions are being used, how pieces connect together. Good docs are notoriously hard to write, and even harder to navigate. But now, being able to query an LLM about specific tasks and get direct references to the relevant documentation sections has been a game-changer.

This realization led me to flip my approach entirely. I’ve started heavily documenting my own development process in markdown files - not for humans, but specifically for LLMs to consume. The key insight is thinking of LLMs as amnesiac junior engineers: they’re capable, but they need to be taught what to do every single time. Success comes from getting the right context into them.

Learning how to craft that context is becoming the critical skill.

It’s not about prompting tricks - it’s about building systematic ways to feed LLMs the information they need.

I’ve built up a library of commands and agents for my Claude Code installation inspired by AgentOS (https://github.com/buildermethods/agent-os) to help engineer the required context.

The tool is a stochastic parrot, you need to feed it the right context to get the right answer. It is very good at what it does but you need to use it to its strengths in order to get value from it.

I find people complaining about LLMs often expect vibe coding to be this magic tool that will build the app for you without thinking, which it unfortunately has been sold as, but the reality is more of a fancy prompt based IDE.

This is why I’m not a fan of auto complete in my editor. Much rather pair program with an agent.

Give the agent as much context as possible and let it go, review and correct the implementation, let it go again, finish it off…

The I just find the autocomplete a little annoying in my workflow, especially with the local self-hosted models I need to use at work.

Claude Code on corporate approved AWS Bedrock account.

I feel like this is the big divide, some people have no use for agents and swear by autocomplete. Others find the autocomplete a little annoying/not that useful and swear by agents.

For me my aha moment came with Claude Code and Sonnet 4. Before that AI coding was more of a novelty than actually useful.