And the internet is largely a platform for capitalism, and most available content is from the internet. Capitalism fundamentally relies on the confidence games of profitable pricing and marketing, not to mention speculation on finance.

LLMs and the entire field of reenforcement learning is fundamentally biased towards the production of Influencing Machines. We are training models at the fundamental level to be subtle and devious con artists.

Yes, I’m just pointing out that phones aren’t terrible at acting as servers, it’s a generational issue.

I’m definitely going to give this a try when I replace my current phone next year.

usbc solved a lot of the connector issues, so long as you can get a hub to play nice with Linux drivers.

Thinking about implementation, it seems like tooltips would be a great way to handle this. Linking out from the tool tips to some kind of more comprehensive outside IT/cybersecurity resource would be a good bonus. Tool tip text generated by llm could take some of the heavy lifting.

Could this be the run-up to Apple acquiring Perplexity? I remain convinced that Apple defending their internal AI division shows they are close to a major acquisition and are just waiting for a valuation dip on one of the major competitors. Distribution is a solved problem for Apple, what they need is proven usecases and a competitive tech stack.

That said, search and consolidating multiple model APIs isn’t a great match for what Apple needs, and their optics aren’t great. My bet is still on Apple acquires Anthropic in 2026.

As a non-coder interested in self hosting and somewhat aware of cybersecurity, this is the most relevant take for me.

An application that facilitates safe self-hosting of many different service is great, however for it to be actually safe and useful it must either be a cybersecurity service keeping up with the pace of threats (which is essentially the corporate closed source model) or from the ground up be an educational platform as much as an application. Documentation needs to not only be comprehensive, but also self-explanitory to a non-technical audience. It is not enough to state that a setting or feature exists, it must also be made clear why it should be used and what the consequences of different configurations are.

This approach is almost never done effectively by FOSS projects unfortunately. Fortunately I think we are at the point where it is completely feasible for this type of educational approach to be fully replicable and adaptable from a creative commons source to the specific content structure of the application user manual using LLMs (local ones). The big question is, what is the trusted commons source of this information? I suppose there are MIT and other top university courses published for open use online that could serve as the source material, but it seems like there is likely a better formatted “IT User Guide Wiki” and “Cybersecurity Risk and Exploit Alert List” with frequent updates out there that I’m not aware of, perhaps the annals of various cybersecurity and IT associations?

Anyway I’m aware this is basically calling for another big FOSS project to build a modular documentation generator, but man would it help a lot of these projects be viable for a wider audience and build a more literate public.

awesome! glad to see it.

https://youtu.be/dU3ru09HTng

Welp, looky there, an expansion port right on the bridge of the headset with PCIE compatibility.

aren’t I glad I just bought an Onn instead.

I don’t see it in the hardware design, but from a software perspective the groundwork is there for modularity. Offloading the core compute to the PC frees up onboard processing to run peripherals like full color front cameras (onboard are black and white / IR) and more advance proximity detection, hell hook up lidar and go nuts with full body tracking.

That said, all of that would depend on decent I/O. 2x USB4 ports would go a long way.

It days right in the marketing text that the headset is “a PC” which to me implies full SteamOS distro with no limitations on installing a different OS, if you can get the many hardware drivers to work.

I remain convinced they have held back budget on AI because they are waiting for the bubble to burst so they can buy one of the bigger developers like Anthropic. Why burn a bunch of cash now just to loose the race when at the end of the day Open Source options might come out competitive or one of the leaders in the space can be bought out once valuations hit a reality check?

Houdini M2 over LORA is kind of a replacement.

YouTube video on it

It’s too bad it leaves the door open for age verification requirements, but the language is overall pretty decent.

Its a commercial product fundamentally. Looking at the company’s site its clear this is an attempt to sell their commercial/enterprise “private cloud” node hardware to the general public but they’ve botched the marketing.

Medical and Transport are their core business, and they are a software-first company that has built a hardware solution for ready drop-in of their secure private cloud server software stack. https://www.nexalta.net/blog-news/11

Looking at NAS options is how I found this, I got suggested a few NAS kickstarters, but the hardware on this one seems to be superior over all. Too bad the documentation sucks.

Well there are 15 days left on the kickstarter but it has been up for a while. I didn’t catch the medical office thing before, but makes perfect sense, they are clearly a commercial/enterprise targetted business and this is their first kickstarter. They just don’t know how to market to the masses.

I agree the software documentation is lacking, they claim it is easy to setup but they don’t show what it is actually like.

I get a sense that this could be a diamond in the rough but to your point about drivers I agree support is going to make or break this device. I think there are some indications that could be decent, the company itself appears to be software-first and targeting highly regulated industries (medical and transport) that require zero downtime. So long as the company itself survives I would guess drivers will likely stay updated. As long as the company survives.

To that point, it seems like this kickstarter is a line in the water for rebranding their enterprise “private cloud” hardware for general use, but they half baked the launch.

IDK, I’m tempted, but without better documentation it’s hard to spend that cash.

What do you think? It isn’t cheap but seems like great hardware to a n00b like myself, I like the future-proofness and repairability of the slots it has. Possibly worth it?

Makes sense to me. I have always thought that if the goal is to emulate human-level intelligence then developers should consider the human brain, which not only has multiple centers of cognition dedicated to different functional operations, but also is massively parallel with mirroring as a fundamental part of the cognitive process. Essentially LLMs are just like the language centers being forced to do the work of the entire brain.

More functional systems will develop a top level information and query routing system with many specialized sub-models, including ongoing learning and integration functions. The mirroring piece is key there, because it allows the cognitive system to keep a “stable” copy of a sub-model in place while the redundant model is modified and tested by the learning and integration models, then cross checking functions between the new and old version to set which one gains “stable version” status for the next round of integration.

Anyway, thanks for sharing.

Thank, try now