Hacker News: danpasca

New comment by danpasca in "AI-generated posters don’t have to be horrible"

danpasca — Sat, 19 Sep 2026 12:06:31 +0000

My point is that those kinds of prompts are dumb and I actually expect the model to produce slop. In programming we write these huge prompts that go into a lot of detail, and when we do that, it actually works. I've never asked a coding agent to go beyond the mean, that just sounds like a dumb thing to do. Apply this rule in design and I'm sure you'll get better results. Kinda what the blog post is all about.

New comment by danpasca in "AI-generated posters don’t have to be horrible"

danpasca — Sat, 19 Sep 2026 11:14:06 +0000

> But AI has absolutely no problem with going with the cheesiest, most overused trope.

Considering your prompt fits exactly that description I think the AI delivered perfectly. Try using some actual design language and see how well it does.

New comment by danpasca in "AI-generated posters don’t have to be horrible"

danpasca — Sat, 19 Sep 2026 11:08:21 +0000

It doesn't fail, it actually demonstrates how the model can improve the output if the prompt is better. AI generated websites look like crap because developers don't know any design language so they have no idea how to prompt the models to get something good and more unique and creative out of them. This shows that you can actually achieve that with just slightly better prompting. Just like with programming, actually.

New comment by danpasca in "OpenScreen is an open-source alternative to Screen Studio"

danpasca — Sun, 05 Apr 2026 00:58:14 +0000

Yep. I've had ChatGPT build many simple python scripts or browser extensions that I can use for free instead of paying for subscriptions that shouldn't be subscriptions in the first place.

New comment by danpasca in "Universal Claude.md – cut Claude output tokens by 63%"

danpasca — Tue, 31 Mar 2026 03:07:30 +0000

The file starts with:

> Answer is always line 1. Reasoning comes after, never before.

> No explaining what you are about to do. Just do it.

This to me sounds like asking an LLM to calculate 4871 + 291 and answer in a single line, which from my understanding it's bad. But I haven't tested his prompt so it might work. That's why I said be aware of this behavior.

New comment by danpasca in "Universal Claude.md – cut Claude output tokens"

danpasca — Tue, 31 Mar 2026 02:39:30 +0000

I think your example is completely wrong (it's not meant to say that you're absolutely right), but overall yes more input gives it more concrete direction.

New comment by danpasca in "Universal Claude.md – cut Claude output tokens"

danpasca — Tue, 31 Mar 2026 02:11:06 +0000

I might be wrong but based on the videos I've watched from Karpathy, this would, generally, make the model worse. I'm thinking of the math examples (why can't chatGPT do math?) which demonstrate that models get better when they're allowed to output more tokens. So be aware I guess.

New comment by danpasca in "Apple Business"

danpasca — Tue, 24 Mar 2026 22:44:31 +0000

> Starting April 14, Apple Business will be available as a free service in the U.S. and 200+ countries and regions to new and existing users of Apple Business Connect, Apple Business Essentials, and Apple Business Manager.

> Apple Business Essentials, Apple Business Manager, and Apple Business Connect will no longer be available once Apple Business launches. Business Essentials customers will no longer be charged their monthly service fee for device management after April 14. Existing Business Connect data — including claimed locations, place card information, photos, organization information, account details, and more — will automatically migrate to Apple Business at launch.

I don't get it. Is this free? If so this is insane value compared to everything else.