I love these types of thing. I'm curious what else all folks are using. Agent-of-empires seems pretty popular, has nice web and phone. Tmux based, which I dig. https://github.com/njbrake/agent-of-empires
It's not great but I made a typescript library to wrap pickers recently, such as skim, fuzzel, fzf, dmenu, rofi, etc. Some very similar problems.
Would love if anyone has thoughts or suggestions. It was quick and dirty, and works fine for my use, but I'm not sure where else I could take this, how else I might splice apart the problem, what else would suit it. https://tangled.org/jauntywk.bsky.social/picker-power
> are extremely powerful when used with deliberate care. Gas Town is the exact opposite of what is needed to actually do useful things in prod.
This judgement feels premature.
We really don't have any idea what is possible. The wheels within wheels, epicycles within epicycles model of agentic loops hasn't really been deeply explored and we just don't know where they might go.
I too share your instinct that human steering helps, helps a lot. But I've found I can keep using less of it, as I setup better parameters, as I improve conducing the LLM into good paths. The idea that an LLM could sit up top and help try not one idea at a time but try many things, then pick and cobble together next goes: that is madly madly madly exciting to me.
I don't want to keep being the bandwidth limiter in this system: I want to scale out. I haven't been following close or trying but I tend to think while total hands off is not the way, having LLMs that can cover a lot of terrain, explore a lot of solution spaces and directions, then assess how to put it together & what to take forward, and other practices of agents watching agents, has enormous potential.
Deliberate care relies on pre-obtained wisdom, and often, the human biases kind of suck and aren't they good. We aren't great at burning down our systems enough, at Chad Fowler Phoenix Architectures. I think the AI's lack of over deliberation and it's ability to try vastly more could be a huge advantage, could show diversity triumphing over specific crafted intent.
It must really really suck to be a data-holder, that every single government out there views as some piggy bank, sitting there waiting to smash & grab.
It's certainly been quite the turn recently. But being between the people and the governments that seemingly inevitably will turn into arch fascist pricks & go to war against the citizens is not an enviable position. Hopefully many jurisdictions start enacting laws that insist companies build unbreakable backdoorless crypto. Hopefully we see legislation that is the exact opposite of chat control mandatory backdoors. It's clear the legal firewalls are ephemeral, can crumble, given circumstances and time. We need a more resolute force to protect the people: we need the mathematicians/cryptographers!
I feel like there's another component: that the consumer base has become so detached from making things in general. We are surrounded in ever more stuff, ever more material, but collectively are out of touch with making things, with material, and assemblage there-of.
Our culture's perspective is as critic, as shopper, as buyer. Sure few of us were expect shoemakers or backpack makers, but people around us were industrious, did provide labor to make goods that people around them bought. The cycle of production had been directly apparent.
This is low key one of the things I really had hope for for a while with 3d printers: that they opened up & exposed what is. That they would be a force to spread insight & to regard the little mechanisms and means of the world all around us. I think that's a little bit true, but it's pretty niche, and I expect most prints are for static parts; no movement or dynamic behavior. And it's somewhat the anti-process: crafter in a box. It's still amazing but barring major changes, I have over indexed.
It's also worth noting the role of DMCA anti-cirumvention laws in casting mankind out of ever coming to grasp with what makes up the world. The combined legal and technological destruction of any right to repair is really not just about repair: it's an obstruction to humans understanding the world around them. We cannot become savvy in the world when the government tells us that business's right to keep us from knowing the world outstrips any mankind-the-toolmaker / natural scientist role/title/god-bourne nature, that cutting us off from the universe & living in ignorance is a hard cast legal binding fact. I find this to be as fallen as it comes. How do we stay alive as the race we were when our laws unwind the fantastic graces of inquiry the gods saw fit to give us?
It could be a cli tool, and it should be a cli tool, for exactly this reason.
Let the LLM work in code mode. Don't make it have to be the execution engine too. It can do it but it's slow and giving it tools script what it wants will go far better.
I do think there's an interesting possibility where we turn MCP into something composable. Capnproto has promise pipelining where you can issue new instructions with results you don't have yet. If MCP could copy those tricks, & express promises... and those promises worked across MCP servers ("third party handoff", https://github.com/capnproto/go-capnp/issues/597)... you'd start to have something as compellingly composable as the shell.
DECT-2020 aka DECT-NR+ looks cool. Basically just used the name, is how I once heard it explained, with really nice modern low latency communication.
https://en.wikipedia.org/wiki/DECT#DECT-2020
And the spectrum has a chunk of bandwidth, just for DECT! That's used less and less. That this new modern protocol can use! Neat.
I haven't bought it but I do keep thinking I might try getting an nRF9161 devkit or whatnot that has support for for dect-nr+. I want to build a wireless mouse with it.
I gave Jellyfin up and went back to upnp/dlna after the Android and iOS clients would keep losing sync, or wouldn't show me some season of a show, or would pick a white background on white text for a show.
The pain just kept adding up. It was quite nice most of the time. But every single time I reached for my phone, I was wondering how badly it was going to go. Quitting Jellyfin seemed like an excellent choice.
Upnp/dlna is much cruder; very direct raw BubbleUPnP client. But it works so well for me. Their transcoding server also is quite good and I can run it on any machine I want, isn't coupled to anything, can switch between them easily.
Bubbleupnp is also great because it lets me turn tablets into cast screens. I love that so much. Good general protocols rock; having media server, media renderer, then separate control points was a great model, good job UPnP.
reply