nusl

I have the feeling that this is going to open the floodgates for persistent autonomous agents, moreso than what's already been happening. Similar to when Apple does something that was already being done. Let's see.

Do models even know their own weights to be able to do this?

No, I've received it just from entering my e-mail and nothing else. So many companies don't care, or don't know.

Either way, if I enter my e-mail, decide "nah" and quit, that should be that. It shouldn't be sending my details in the background

Shopify does this. If you ever go to purchase a product on a Shopify 'site, and enter eg your e-mail address, it'll save it even if you decide to cancel and close the tab. Sometimes you receive spam from companies along the lines of marketing or "we noticed you didn't complete your purchase" type crap. Shit is rly invasive.

May be possible to repair the existing one?

Yeah, because it's 17% when compared to the current +50% "promotion" they have going. They're raising base weekly limit by 25%. It's still more than non-promotion usage limits.

Fair. I'll rm the personal project bit

I hope we see fewer comments like this that aren't shitting on projects that you have no stake in anyway

This is super cute. I haven't tried it or anything but it's really fun, seems genuinely useful too. I don't really get people calling it cringe.

"written by a human, on purpose."

idk about that

Haha, fair enough. I read the Dune books through the 4th and enjoyed them. I saw some typos and things and figured maybe there's something hidden in there.

Yeah, maybe. The issue I foresee is more that the moat is quite small if you can just build it, so the moat would need to exist in stuff that can't just be built and replicated within a few weeks or something. User base maybe, connections to real resources, some other form of deeper value you can't just copy. Not sure though; I'm not a business person

Is the Dune summary in your bio steganography of some sort, or contains a code? Or just as-is to avoid spoilers?

Suppose you could use voice too, yeah. I built a k8s-based sandbox environment. One pod = one agent, and then a bunch of scaffolding around that to manage it and stuff. "Agent" here is a Claude Code instance within a pod with root access to do whatever it wants inside the pod. The main interface is Discord, but it supports other interfaces/frontends. 1 agent = 1 pod = 1 discord channel, managed by the harness. It can do other stuff too, like spawn its own pods and namespaces to build and host its own infra, new agent teams/sub-teams, but unrelated.

I mostly use text to talk to it, and send files via the Discord file sharing mechanism. Discord has voice recordings too, which are sent as .ogg files, which the agents then turn into text with Whisper.

Not quite plain Claude Code but I didn't want to go into too much depth in my original message.

The end result though is a persistent data store for each pod/agent, and then I can talk to them whenever I want from wherever I want, and interact with it with proper media support and stuff. So I have one agent that's for medical logs and such, and others for various other things.

I have an llm agent with a harness that stores/logs everything in a structured manner to local files and/or sqlite. This then turns more into casual conversation type stuff. Send it test results, talk about stuff, give it updates, etc, and it handles the rest. Querying the data is pretty straight-forward without the agent since it's plaintext/sqlite. I tried random tools over time but sorta fell out of them. This is kinda the only thing that's stuck.

Could take this further and make it better with structured tooling. Some sorta CLI perhaps, or an API, that the agent knows how to use, then can build a frontend on top or something. For now this works okay.

Anything with Tencent on it is a no for me.

How is this a conflict of interest instead of a plain naming conflict?

Related source-source: https://not-an-llm.com/meat-based-llm-proxies

Reframing the thing you’re doing as “trying to do the thing” struck me as something kinda powerful. You’ve got to suck at something before you don’t.

I’ve found that if I start something new, say learning to carve wood, I might compare myself to others in an unhealthy way, not improve as fast as I want to, or something else, and sorta psych myself out of it. I wonder if changing the framing from “doing” to “trying” would help

This is really cute but also really funny.

I've built something quite similar, actually. It's quite a powerful workflow. Integrating multiple agents into chat with freeform comms is pretty fun, but I do find that you lose fidelity of what the agents are doing in terms of seeing actual changes, tool uses, and other things. The moat here does seem quite small, though.

This is genuinely really cool. I guess the added complexity and/or performance hit will require some justification but I can see it paying off for some use cases.

Okay, I stand corrected. Apologies for making the assumptions I did.

In the age on LLMs it's definitely way easier than what you did.

I think copying a website like this is very poor taste regardless. If I see you doing this, I immediately lose trust in your product and will immediately leave.

If you can't put the effort into the face of your product, how can I trust you to put effort into the product itself? Shitty behavior, with a shitty justification self-affirmation blogpost.

Terrible fix but it's a fix that's minimally-invasive and addresses a bug that causes a disproportionate annoyance to the fix. I can imagine your cursor lagging is something that is extremely annoying over time.

Wonder why they chose to stick with the portafilter form factor rather than something custom for the method. Perhaps it just fits the task well and they don't need to change, or they want to preserve the look of an espresso machine. Though, if the water is room temp, they only really need the pressure, which doesn't need the entire machine. Seems pretty cool.

Edit: The article shows that they chose this method specifically to use the basket, and the basket allows for the vibrations they need. The other conveniences of the basket form factor also come into play; reloading, cleaning, etc. and don't need reinventing or retraining. Cool stuff. I wonder whether they'll remodel the machine around the basket and remove what's not needed if this becomes a product.

I've seen this happen a bunch too, though fortunately it hasn't been _that_ common. More often is managers that don't understand things using AI tools to try to understand them, mostly failing, and then regurgitating the LLM output during a meeting. Added as a link on my blog, too, since I have a similar article.

Monthly subscriptions for software like this confuse me. The website also obscures that it's paid at all - the first option is "download for X", then the next option is a red button to "unlock". No pricing link/page, nothing. Aimed at getting you to download and use it, then fork up later.

References mentioned all going on about productivity etc. Red flags all-round.