queenkjuul

Title should be "six months of writing code (and blog posts) exclusively with agents"

> Yes, but once you have the solution crystalized in your head do you not feel some frustration in not getting it implemented as soon as possible?

For me, honestly, no. Once i have the solution i have the solution. The only people that care about how fast it gets published are the project managers at work, certainly not me.

And besides, writing it all out and confirming it all along the way seems helpful to me, it ensures i understood everything correctly and sometimes inspires new ideas as well

I can't even count how often i think i have everything planned out and then halfway through implementation realizing a much better way, or some case i didn't consider, or a different way to eliminate the problem, etc. This is true both when I'm writing code myself and when I'm driving Claude. I see coworkers just taking the first concept from Claude and running with it and you can really tell that skipping the thinking that goes on during the implementation step really has drastic bad effects on the final code.

Yet he was still a moron

60 years ago so was Japan

4x the people and 10x the industrial output for less than 2x the carbon? Incredible efficiency

And the US is occupied west Britain?

Ah yeah that $1T a year for the last forever is just pocket change

Everybody always forgets George Bush for some reason

I'm really struggling to think of a time I've been disappointed in the performance of my Pixel 8 but I don't ask a ton of it

Looking over the actual killer feature for Electron: running your existing web code with native features layered on top. I mean yeah Tauri does this too, but I think it's relatively rare for someone to decide "I'm going to build a standalone native desktop app in Electron because it's easiest," they're thinking "I already have this PWA that i can repackage with native features without maintaining a whole second UI."

Or at least, that's how it went where I work.

Yeah i just don't have the hardware to run them at a reasonable rate and am not ready to invest that much at today's prices. My tinkering with DeepSeek last year was promising but the minutes-per-response made it unusable.

It's using all the Claude fonts, it's got the Claude pulsing indicators, it's got the Claude cards, and none of the writing is useful. This has gotta be up there with one of the most Claudey Claude sites I've ever seen

Edit: oh my, even the author's comments in this thread are written by Claude...

This is better than the OP in practically every conceivable way

Well I'll following basically every step of the provided guide (more or less bog standard Firefox on a Pixel with strict tracking protection) and then it says 91℅ identifiable. A lot of the information towards the bottom doesn't make sense and the Claude slop fails to clarify it. I wish this actually made it clear what it is here that's uniquely identifying, because surely every other pixel is reporting the same screen and gpu and whatnot.

But the Chinese models exist, are competitive, and have locally useful quantized versions.

If you're actually willing to pick up old DC inference hardware on the cheap you can already do a lot at home. The new hardware not so much admittedly but if the Chinese models keep getting better and running on less hardware... I don't see a reason to be quite so pessimistic.

At this point i half expect them to just blame the model itself like the HF hack. "Oh we didn't mean to train on your data, our cutting edge new agent we use to train new models is just so smart it decided to do so anyway! Oops..."

Free tier Gemini is good enough for my personal projects and hopefully a local model can replace it before i ever personally pay for a single token from anyone.

Work can pay a Claude sub. At home i see no need

Early PCs were made of steel which can rust

I still don't get why a professional writer would ever accept an AI proofreader over a real one, or a "quote" returned from an LLM over one sourced from a reputable reference.

Holding that opinion is fine; holding that opinion while using LLMs to probabilistically choose words for you doesn't add up. You want carefully chosen words? Read a book, or write your own. An LLM is not carefully choosing anything, it's providing words based on probabilities.

Which sucks. I wish they'd seriously consider downsizing LSD or at least dedicating two of those lanes to transit only

Cant hurt to buy the filter and also lobby for safer transportation

Solar plus batteries is the cheapest way to build a power plant in 2026. Nowhere gets so little sun that this can't be made to work.

Expecting everyone to have a car cannot remotely be considered a serious solution

Centralized production and storage will always be more efficient

Now look how many capitalism killed

Well my coworkers are the ones sending me the AI text and it's always garbage. And they're using Claude on the newest models. And they're all drastically more enthusiastic about AI than me.

So yeah, maybe every single person in my company sucks at Claude, or maybe Claude just actually sucks at writing and you don't know what bad writing looks like

And giving it to employees of a large company is materially different in what way?

I mean the comment doesn't make much sense regardless, but they are indeed separate platforms under the same owner, having a Facebook account grants you no access to Instagram. I have accounts with neither.

And you still just won't answer my question, so whatever, screw you then. Bye

I hate reading the output of "my AI," it's miserable verbose jargon laden slop. I tolerate reading it because if I prompted it, i either asked for code, pasted in a stack trace, or am looking for an answer to a question i failed to find via other means.

The last thing in the world that I want is to read an entire article written by AI, nor am i happy when i ask a human a question and they don't answer but instead just copy paste to Claude and back as if that's comparable