I seem to spend half my life logging into thing's, confirming 2fa,confirming biometric data. Then when I go back to the first thing it's timed out and I have to sign in again.

Very interesting! How does it work?

The main worker is a .Net API with Hangfire that hits a public facing API from the radio station. That data is stored in a MySQL database that feeds a React dashboard front end.

There's some caching that happens behind the scenes to help with some of the large data crunching.

Ah the radio station has an api. I was thinking you were doing song matching or something, nope much simpler than that!

I started something similar but for Sydney with the ultimate goal of being an agent simulator, simulating everyone going to work etc. Project got a bit unwieldy and is on hold while I mull it's future over.

Quarries were great but I hated the big hole it left in the ground.

Because the reviewer ends up doing the real work actually checking it works.

The laziness is offloading work down the line.

That has nothing to do with using AI, if the dev didn't check their work then that is being a bad dev.

That’s what this whole thread is about. Appearances of productivity, laziness, and the offloading of real work downstream by sending of “looks good enough” ai generated work.

The tears of sysadmins are fairly cheap though.

It may be the only resource that truly scales with demand.

And how much does the hardware cost to run said models?

How good do you want it to be? For a close to ChatGPT today (April, 2026), you're still looking at a system with 7xH200+chassis, which will run you $300, or a GB200 NV72, which is $2-3 million. OTOH, a Qwen3.6 quantized model can be run on $10,000 (high end Mac) or $1,000 (Mac mini) worth of hardware. Even a Pixel 10 Pro cellphone ($1,000) can run useful models locally.

Go to Open Router, ask your own in investigative prompt that meets your needs to all the top open models. See how they do. Then notice if you can run any of those locally. Repeat at least once a month.

Thanks, BTW, now I have learned about OpenRouter.

It doesn't look like they have a way to filter down to "open" models. By this of course I mean "downloadable, local models".

I suppose if you know the "family" (Gemma, Qwen, etc.), I can just go to those models and test…

I've simply been pulling down what is popular from the LM Studio front end (and what runs on my hardware) and testing in situ.

It can be quite expensive to get the models and machines to do this.

That's what the money pays for when the Comment above mentions 'that you might have to eventually pay an AI company a large amount of money to ask ChatGPT such a question'

Putting aside that it won't be a large amount of money For any particular query , that's how the AI companies see themselves, not as providers of information, but as providers of mechanisms that provide information. It is not selling the Information of others, it isn't selling information at all. They are selling the service of running the mechanism.

You can run them slowly on any machine that has enough memory.

And, to bolster your comment, you can still use this machine as your daily driver.

I'm always going to have a machine anyway—might as well max out the RAM when I purchase another.

(And so too I jumped on the Mac mini bandwagon a month or two back—64 GB. I'm enjoying pulling down the new models and putting them through my paces.)

Because cursor gives you access to tons of different models, not just the Claude models.

This is outdated,

Claude code can be run against any model you want, you just simply need to update the ENV vars.

Or if you have Ollama running locally you can do something like this:

ollama launch claude --model glm-5:cloud

Or if you feel brave and want to turn your laptop into BBQ

ollama launch claude --model minimax:2.5

or you could just use cursor and select between everything at will every request and not have to manage 15 subscriptions

Exactly that. I don't want lots of subscriptions. Zed with OpenRouter sounds all right except it'll be much more expensive than a subscription, I fear.

Well fuck that then

There are still BBS you can access via telnet (and actual dial up if you really want), after the fifth one asks you for your full name, street address and phone Humber it gets a little old.