sky2224

What questions in particular are you thinking of that's purely about the engineering and ultimately not the political agenda?

Why did you need Mythos to do this? There's nothing that seems inherently difficult about what has been built here.

Correct me if I'm wrong, but it seems like you've built a fancy looking site for a service that essentially generates a prompt for something that may be related to your business, and throws it into each of the major AI services (chatgpt, claude, gemini, etc) to see if it name drops your brand... and you want to charge $49/mo for that?

If you can get someone to pay you that, more power to you, but sheesh.

...why? This is something you could automate in about an hour using your own LLM subscription to drive the prompts.

Just mentioning this since no one else has yet: it could be your tonsils and/or adenoids, so it may also be worth seeing an ENT if you suspect this is the case.

Cause driving is fun and these systems get in the way without providing any noticeable benefit to me.

I tried lane keep assist in an audi a5 sportback, and it was genuinely terrifying to me. It felt like it was trying to drag me into oncoming traffic when going around curves.

Automatic braking sounds great on paper but practically speaking doesn't work out in my opinion, and once again seems to get in the way. Many people have reported getting frequent alerts while driving or even having ghost braking going on with their Subaru EyeSight systems. This wouldn't be an issue if I could confidently remove the system from my vehicle, but right now I practically can't without doing some heavy lifting on my own.

That definitely addresses the privacy issue, but some of the automatic braking and driver awareness monitoring is a bit much for me. For example, Subaru now puts a camera in the dash that frequently scans your face and makes a beeping noise if it thinks you're not paying attention. As you can imagine, it beeps a lot.

You made a good point that I didn't think through fully. It's the concurrent user aspect that heavily impacts things. Currently, you'd probably need quite a bit more investment to the point of having a mini data center to do what I'm proposing.

However, we've been seeing advancements in compressing context and capabilities of smaller models that I don't think it'd be too far off to see something like what I'm talking about within the next 5 years.

This is part of the reason why I'm really worried that this is all going to result in a greater economic collapse than I think people are realizing.

I think companies that are shelling out the money for these enterprise accounts could honestly just buy some H100 GPUs and host the models themselves on premises. Github CoPilot enterprise charges $40 per user per month (this can vary depending on your plan of course), but at this price for 1000 users that comes out to $480,000 a year. Maybe I'm missing something, but that's roughly what you're going to be spending to get a full fledged hosting setup for LLMs.

Honestly, I think part of the reason Apple hasn't jumped deep into AI is due to two big reasons:

1) Apple is not a data company.

2) Apple hasn't found a compelling, intuitive, and most of all, consistent, user experience for AI yet.

Regarding point 2: I haven't seen anyone share a hands down improved UX for a user driven product outside of something that is a variation of a chat bot. Even the main AI players can't advertise anything more than, "have AI plan your vacation".

It's probably one of the biggest headlines right now. OpenAI has about $96 billion in debt and they don't have a revenue generating product yet.

No, I don't think so. You've paid for a service that will run an AI model given some prompt. There have been zero guarantees made that it will actually solve your problem.

As others have stated too, how do you define what an incorrect output is?

Can you provide examples of YC startups that knowingly broke laws and just dealt with those issues later? I'm not very aware.

This is a sales post in disguise.

03 audio sourced from the web

Where? How do I know you're not pulling from some shady repository?

Chegg is a service many students used to get guidance and answers to homework problems for whatever courses they were taking. It was a sinking ship once GPT 4 came out, but GPT 5 was really it's final nail in the coffin.

I don't know any student that really uses it now.

No kidding, it took my CPU usage from 1% to 55% instantly sheesh

I am suspicious of grifters and would like to find trustworthy advice.

If you want actual good advice: go to a doctor.

Don't go to a chiropractor, don't go to hackernews. Go to a doctor. You can either start with a physical therapist in your area or start with your primary care doctor to get a referral.

I'm assuming you're in the US, so I know it's expensive but this will genuinely shorten your life span if you let it get significantly worse.

I feel like people genuinely don't understand what vibe coding means.

Just cause you're using an LLM doesn't mean you're "vibe coding".

I regularly use LLMs at work, but I don't "vibe-code", which is where you're just saying garbage to the model and blindly clicking accept on whatever is spit out from it.

I design, think about architecture, write out all of my thoughts, expected example inputs, expected example outputs, etc. I write out pretty extensive prompts that capture all of that, and then request for an improved prompt. I review that improved prompt to make sure it aligns with the requirements I've gathered.

I read the output like I'm doing a deep code review, and if I don't understand some code I make sure to figure it out before moving forward. I make sure that the change set is within the scope of the problem I'm trying to solve.

Excluding the pieces that augment the workflow, this is all the same stuff you would normally do. You're an engineer solving problems and that domain you do it in happens to involve software and computers.

Writing out code has always been a means to an end. The productivity gains if you actually give LLMs a shot and learn to use the tools are real. So yes, pretty soon it's going to become expected from most places that you use the tools. The same way you've been expected to use a specific language, framework, or any other tool that greatly improves productivity.

I wasn't implying that clever prompting needed to be used. I'm just trying to confirm that the person I was replying to isn't just saying what essentially amounts to "build me X".

When I write my prompts, I literally write an essay. I lay constraints, design choices, examples, etc. If I already have a ticket that lays out the introduction, design considerations, acceptance criteria and other important information, then I'll include that as well. I then take the prompt I've written and I request for the model to improve the prompt. I'll also try to include the most important bits at the end since right now models seem to focus more on things referenced at the end of a prompt rather than at the beginning.

Once I do get output, I then review each piece of generated code as if I'm doing an in-depth code review.

Can you provide an example of how you actually prompt AI models? I get the feeling the difference among everyone's experiences has to do with prompting and expectation.

Weird. I just went to the shorts page now and it says what I described.

I just tried on Chrome and Firefox on desktop, and the iOS YouTube app. Both show me a message saying "Recommendations are off. Your watch history is off, and we rely on watch history to tailor your Shorts feed. You can change your setting at any time, or try searching for Shorts instead."

I'll also clarify, sometimes if you click on a shorts video that you searched for manually, a few related videos will be queued, but then the feed will try up and the watch history message will display again.

Do you have left over watch history from years ago you've never cleared, and maybe shorts is enabled since it uses that...?

fwiw regarding getting sucked into youtube shorts: if you turn off your watch history youtube refuses to let shorts work. It will literally say, "turn on your watch history to continue with shorts".

Even if you told it not to do this, it likely just searched the web and pulled this from github.

If you actually wanted to test this, you'd need to run Claude Opus 4.6 locally, which is not really possible.

I've spent a good amount of time playing Downwell.

The metal slug games are also on the app store. It's a bit hard to play but you can set infinite lives and it's a fun way to kill some time while waiting around somewhere.

What are the numbers on companies indirectly use Claude via Github Copilot? Given that OpenAI and by extension, Microsoft, has folded to the will of the US Govt, I suspect that all Claude models have a pretty non-zero chance of being pulled from Copilot very soon.

I know many companies using Copilot have no contract work with the government, but this is the Trump admin we're talking about. They want to send a message.

I'm a little bit confused about the development workflow with this. It seems like you've developed a system for the AI model to essentially emit steps that it can follow to interact and return information about the state of your app's UI, but are these steps created saved anywhere? How do you ensure consistency for the same prompt submitted twice? How does this fit in deployment pipelines?

Also, maybe I'm misunderstanding what this library is for, but is this really a good application of AI? I'm getting reminded of gherkin + cypress but without the test actually being embedded in the code. I feel like I'd rather use AI to write the BDD test using gherkin rather than prompt the AI to figure out what test behavior I actually want.

I realize I'm coming off a bit pessimistic here, but I'm just trying to explain where my thoughts are so you can hopefully clarify things a bit for me here because I feel like I'm missing something about what you're trying to accomplish with this project.

There are an increasing number of AI generated posts that are automatically posted without human oversight now. Sadly, it's gotten to the point where honestly we truly don't know what is and isn't real, and OpenAI really ruined the em-dash by making it directly associated with AI generation.

Your em-dashes make me think this is an AI generated post but whatever.

My company uses Github Copilot. We have a very specific enterprise agreement that states that data does go to Microsoft's servers where it gets processed in an ephemeral environment and wiped after 3 months.

I'm guessing Anthropic has something similar in their agreements. Now, if you have some proof that Anthropic is stealing highly confidential and/or trade secrets, that'd be good to see, but also whomever is throwing that kind of information into an off-premises and non airgapped model is just asking for a data leak.

There's only so much you can do to detect and block content that's AI generated. At the end of the day, the content starts with the people creating it.

Jumping to an invite only network isn't the most ridiculous idea imo.

Have you spoken to anyone about this in person or on a social app like Discord? Could be a coincidence of timing, but also I have found recently that ads have been significantly more blatant and aggressive in their targeting. I've gotten sponsored Instagram posts for things that aren't really a product like a news article covering something I was discussing with people in a Discord voice chat.