lbhdc

I was on the implementation end of a considerable amount of industrial automation and technological advancement about 10 years ago. When we were on site the result of making mistakes started at the death of a team member. There were a plethora of things that could kill you horribly, falls, hazardous environment, rotating equipment, etc.

Yet we all survived overhauling processes in hundreds of plants. Working in hazardous environments isn't untenable, or even particularly difficult to do safely. In fact we worked at a much faster pace (with fewer mistakes) than corporate world I work in now.

Semantic search to bring up the exact relevant pages is a killer feature.

Having an AI that has all the site specific information in it would also be killer for contractors who are constantly in and out of plants. Everyone has site specific rules, and you can easily burn up a few peoples day buy bringing them onsite and answering their first round of questions. An AI could let a lot of that be self serve.

The reason is spam. Before these can get wide spread "normal" adoption they can be heavily used by spammers. Its hard to say if that is because they have desirable look-a-likes available, or if its because the first year is offered at a deep discount. So, systems will get flooded, and on inspection they will see that they don't have any legit traffic from those tlds and will whole sale block them.

.xyz is kind of infamous for being in this situation. https://news.ycombinator.com/item?id=28554400

I have no idea if that applies to .coop though.

Don't they already offer this? Did it get canceled?

We have an intern from an Ivy (not MIT) that isn't getting an offer simply because my company doesn't want to hire in the US right now. They are great to work with, knowledgeable, but have no future here. They have been shopping around, and a lot of people on the team have been trying to find a place for them to go in their network, but no one is biting.

What if the farmers don't have a reason drive instrumented equipment in an area they want data? Are they now burning fuel and time to gather that?

Is the data gathered that much better than what the human driving the equipment can garner that it is worth it?

One nice thing about this new business model is that it doesn't rely on display ads (not that they can't do both). Conceivably ad rev could drop so low it isn't worth compromising their UX. So theoretically we could get far more readable news websites.

The new friction will be convincing the gate keepers you are a human.

Global news seems to be oversaturated. There are dozens of outlets repackaging snippets from reuters and elsewhere into articles. They come out around the same time, hit more or less the same beats, and for the reader are essentially fungible with other articles on the topic.

Before AI overviews they would use SEO to game their way into traffic. That allowed everyone to publish the same content with lots of friction and the biggest brands would still do well. It seems AI overviews has shown these companies users "revealed preference", and it isn't for their high friction websites.

My guess is whatever is involved is related to the business they are actually in, and (if successful) this is a high profile way to demonstrate their expertise at solving problems in this space (and getting a lot more funding for future swings).

It seems they are focusing on birds, so maybe something like making more resilient chickens for farmers. Perhaps their pitch might be something like spending X% more on Colossals modified chickens yields cohorts that have Y% better survival rate reducing veterinary costs and increasing product to market while assuming the extra cost doesn't overshadow this benefit. Like GMO seeds but for livestock.

There is a lot of talk about "browser agents" vs "recursive crawlers", but that discussion misses what site owners want.

Products that use "browser agents" often dis-intermediate the sites they scrape even if they link back. While users may prefer those user experiences, content owners will pay to block that to retain control over their presentation and monetization friction.

Many news sites already have these sorts of captchas, proof of work, behavioral analysis roadblocks. Players like Cloudflare and Tollbit both offer bot blocking proxies as well. This tech is here and service owners want it.

What audience do you think Digg is missing out on by not having federation?

You can also use duty cycle metrics to scale down your gpu workloads to get rid of some of the slack.

I am sure there are programs that are a mix between a typical BFA program and a teaching degree, but that is not the norm.

I know people will push back and say that is not the point of the university. But it doesn't change the fact that our economy is not built on poetry and painting, but we educate large number of people to specialize as one. Those people are instead left in debt with no path forward in their chosen field.

I get ~20-30 tok/sec. It's fast enough that its not frustrating, but if it were faster you could more easily skim as it generates.

I haven't done any serious testing. My process is typically learning about new models on HN or elsewhere, and trying to give them a real shake. I have some goto code generation prompts that I try on all of them. None succeed but they are getting close. I also do a lot of just feeling it out. The more I can use solutions unedited the better it feels.

Currently I am using llama.cpp for an interactive repl chat. I was previously using Alpaca (a GTK GUI), but was annoyed with how slow it was and some random crashes. I am transitioning some of this to self hosted in the cloud for things that can't run on my laptop.

I am looking to get away from my current interface, and write my own. Mostly for experience of deeply integrating agents into a program. If anyone knows a good library for interacting with a local model that doesn't involve standing up a webserver I am interested :)

My daily driver is gemma3n. Its been a nice balance between speed and performance without spinning up my laptop fans.

I am super interested in local models, partially because there is no friction from managed services, but also because I think as small models become more viable we will see an explosion of apps incorporating them.

Collecting and forwarding the emails to an address / dataset could be beneficial as well.

It created a landing page and a questionnaire to funnel people into email collection.

The landing page looked the best, a gradient with some pretty funny/clever copy related to my prompt. If I was making this for real, I would want more. Some images, more copy about my products, maybe even some animations.

The questionnaire pages felt plain, and could use some styling. It also felt like there were too many, and would probably see an engagement drop off.

The app gives some basic styling options that it does automatically as well as ways to adjust the copy.

The app also provides a way to deploy this, but I think I would want the underlying JS so that I could tailor this to my needs, effectively using it to get over the "blank sheet" problem.

I know the author is accepting patches, but even learning that their forge exists, and that the user wants to contribute is stymied if the users search for projects starts at a github search.