loveparade

Like many other have said, it was probably trained to do this. The authors probably tried other models and it didn't work as well because they weren't trained for this, that's why this article is about Astra specifically. Kind of hard to do science when you are working with a proprietary black box that you don't know how it was built or trained.

Surprised these financials are not too dangerous to release.

All these and many Chinese models are so close in capability anyway that I imagine the price war is gonna continue for a while, good for consumers to get cheap subscription tokens. Will probably be years before the dust settles, and then just switch over to the best value. Something like DeepSeek is more than good enough for most of my work today, just no reason to switch yet while subscriptions tokens are dirt cheap. There is no lock in anyway so no point in worrying about it now. I could literally switch tomorrow and wouldn't miss Anthropic/OpenAI much.

Yeah this is clearly total slop output. could come straight out of "claude write post about how ai can write more code but we have to do system engineering"

It was too dangerous to not release

A large part of software and websites I used before have been replaced by talking to LLMs. Just today I did some video editing, I would previously used OBS studio, but now I just prompt Claude with what I want, give me 10 options to pick from. Then my IDE, I don't use it anymore. Diagramming tools? Nope, prompting works better. Image editing? Prompts. Web design tools? Nope, prompts and give me 20 live previews to iterate on and narrow down. Web consoles to manage my servers? Nope. Excel spreadsheet for workout planning? LLM does calculations now based on my tracker API. I could go on.

So yeah I think we disagree. It feels so much better to me not having to switch between different tools anymore. When I read code it's not in my IDE but PRs on GitHub where I leave feedback for the LLM. I just tell it to keep iterating and monitoring PR comments until I approve.

Agree it feels super early for a lot of these tools. Don't think we have even scratched the surface of what working with LLMs will look like in 10 years. Just think about how many user interfaces can be completely replaced.

For the LLMs so they get new training data

Another reason for why open source models will win in the long term.

You dont have to vibecode and maintain it when there are 20 other projects that do the same or similar thing. And maintenance is not free, but you know waht's even more annoying? Getting locked into a product that is pressured to achieve VC returns and thus becomes enshittified. There is no moat here and it's painfully obvious that open source will win in this category in the long term.

For all applications that support theming you can ask an agent "make this look like windows 98"

don't quite understand how typing it into a popup instead of an agent makes it revolutionary

I guess there is some nuance here. Would I pay for a very good agent multiplexer that makes me a lot more productive? Sure, same reason I pay for other apps or services.

But for Herdr specifically? Probably not. There are tons of similar free alternatives and even vibe coding your own is very easy with all the open source code of both herdr and all the alternatives available.

I really enjoyed herdr at first but ever since they got VC funding I am avoiding it. Enshittification and cloud features due to vc/monetization pressure incoming.

Can I get dividends as tokens?

I don't think so? I have had plenty of claude sessions running for days, just tell them to maintain a markdown file with state and check stuff in the background and react to it. How do you think these PAs are implemented internally? Pretty much like that.

If you want it to answer emails all you need to do is "watch for emails coming in answer them" and it will write a background watcher script for it and manage itself, it's just a prompt. Of course assuming it can access your emails/browser, but that's the same with these personal assistant products.

No idea what you even mean by "a session times out"

This is cool, but I still don't understand how this is different from just a Claude session, with a browser. Everything you described here is, from my experience, easily handled by a regular agent session these days. I fail to see what the personal assistant (dot, etc) adds.

I also built something like this for myself. I call it a markdown file.

Astra is some of the worst code I've seen in terms of slop, so the benchmark is kind of surprising to me. Astra gets things right, but I need Claude clean up its slop all the time. Many times I've cut down Astra changes from 1000+ lines to ~100 lines.

That sounds more like a religion than science. Why not be happy that there is a new technology that helps solve (presumably) important problems that can have positive effects?

Before chasing super intelligence maybe fix how bad the average K12 education is for humans compared to the rest of the developed world.

Would love to see these 10 projects that are being worked on simultaneously

I think it would be more efficient to post on HN when there is a day without GitHub issues.

I think it's a pretty clear line. Self hosted means you can run it locally in your home, using a saas platform or vps is just a convenience, not a requirement.

I'm more worried about TikTok than AI when it comes to future generations.

Done by an internal model that is too dangerous to release.

This. At this point I don't really care about other models because max subscription are super cheap (relatively speaking) and I don't hit my limits. Even if the frontier models are only 5% better I might as well just use the best thing available if the price is reasonable.

Once the subsidization ends and cost becomes significant I will take a serious look around for the best value models and switch off the expensive providers, but that time hasn't come yet.

That's the point. When a human works on a problem they realize themselves "wait, i probably should re-architect now" - of course you can ask an LLM to do that, but at that point you already know yourself what you need, which defeats the point of LLM working on difficult problems that require insight automatically. A lot of these math problems are sessions over many hours. And of course you can also ask "Think about whether to re-architect at each step" and it will never do the right thing because the context it builds up for itself drives it into a specific solution space. It's literally trained to complete exactly that.

I'm a lot more skeptical. I'm not a mathematician, but from my use of LLMs a very clear pattern of what they are good and bad at has emerged. They are extremely good at combining large amounts of information, and it seems this is what the current AI results in mathematics are. There are so many subfieleds of math with ties to each other, so many papers and niche results, that no human could ever read, comprehend, connect and organize that information in their brains. Pretty much all of it was created by humans. And there is real value in doing this and creating new results from what we've already discovered.

But there also is another type of discovery that requires taking a step back and looking at the problem from a different angle. If you are an engineer, how often has an LLM told you (without you explicitly prompting for it): Wait, what you are doing here doesn't really make sense, there exists a much more elegant abstraction that nobody has thought of, let's remove all that code, let's tackle the problem in a different way by thinking from first principles. Pretty much never. But in science a lot of the biggest discoveries have come from this kind of first principle thinking, questioning existing work and approaches and going against what already exists, not combining all existing data which is likely to be just a local optimum.

Hasn't it already transformed the way we live and work? I don't know a single person who isn't regularly using some kind of AI chat product.

That doesn't mean it's not a bubble. Good enough models are commodities now. If OpenAI and Anthropic disappeared tomorrow I would be a bit sad because their models are maybe 10% better than the others, but it wouldn't matter that much really.

Not only could I not care less, but it also makes me want to drop OpenAI for Anthropic. These kind of slop provider lock-in products are exactly what I don't want. Just make a good model that I can use however I want. This is the typical enshittification you usually see at startups when they are desperate to make money to get VC returns, just at a larger scale.