What I am trying to build is the system recently outlined in the ORG card.
It's a field system for environmental monitoring and assistance. It would be nearly 10 years before it could be completely autonomous and having to rely on outside models kinda puts your medical and personal data at risk. The solution I came up with is a Local preparation environment. So, as you sleep It takes your calendar, preferences, operating environment and it does as much research and asset building as it can, so, not matter what, your field device is over prepared for the next day.
I'm a researcher more than a builder and I'm beset by an awesome crippling dyslexia so coding is just brutal for me(I enjoy it and it works but most people say it hurts their brain to even look at) I've plotted the societal inflection points(security concerns and privacy issues and they all point to the local data environment being essential) and I understand the technical foundations of all the different tech that makes up AI. The current Solo-leveling system design would work more so as a base for all future intelligence work outside Huge generalist models (like astra and fable.) The Users knowledge base collected from his sensors and choices/preferences would work to build out a large database like wieghts and token definitions for his system(In a private SQLite table with 4d embedding for fast retrieval and training for his field assistant and so on.)
I realized that the collection of this environmental data, if it can be stored in an append only and provable way, would be valuable to prove real work and experience. Not wanting to share and personal data it would be translated out in a simple explanation by the field devices model(ask-"Did your user clean the park for an hour?" answer-"Yes, approx 1 hour and 30 mins. and meets your requirements" reward- "user can have a free ice cream.")
A system like this, with a user controlled profile(with proven skills, expertise and assets etc) and tools and so on for their day/endeavors with a turn by turn assistant, would make most societal interactions easier. users can host organizations (like a science club/rod and hunters club or explorers guild where they map out an area and it's assets for the guild maps and build tools/quests/rewards etc for whatever is important to them.)
Ive built a few different versions of the different parts. I would like to find some development partners or testers to give some feedback.
so you have a preparation environment--Field device both independent of a cell network. Ideally something with kernal support in a Graphine OS like cell device or a custom watch like deployment because the models are mostly under 50MB for the field device.

I read through everything you wrote in this post, and as someone who has been in the AI field for a number of years now, having used the majority of Open and Closed source models extensively, my honest opinion about this project is that it sound like a slop machine.

This is everything and its nothing. You use fancy language to describe things that have been done 100 times before, and it contributes nothing to the field of AI. I recommend getting your mind out of the AI hype, and actually thinking of practical use cases for AI models, as well as building architecture around those models for systems that can be used in everyday life.

Rather than building the next swarm of Jarvis agents that every other tik tok star has already built and been made fun of by people who actually want to find productive uses for AI.

Like dont get me wrong, i dont want to be a buzzkill, and if you have something you want to create that is genuinely unique and helps solve a problem that has not been solved yet, Im all for it and Id support you in it. But this just seems like everything other project Ive already seen thats only driven by non-sensical buzz/hype.

I understand it absolutely has a wide scope and it is Very new tech so I respect your skepticism. The LLM is definitely not an integral part outside of translating the data to language. Not 100% where the slop would come from. The Prep Environment? I've built one out already which outputs an "Endeavor package". so whatever is on your calendar gets research and assets built out for it. There is a Captains log at the end of the day that does a few things but mainly it builds preferences and gives feedback.

Do you Know what ATAK is? Think ATAK with heavy user preparation for a "Endeavor package" (Has Profiles of people from publicly available info or past interaction. It traces and places assets which could be relevant based on user preferances)

The actual field device/application is a few parts. User Profile, An append only evidence system. Environmental sensor effector cascades for monitoring which reports to the append only memory. Mainly it is a platform for people to use AI and custom tools in thier daily life without having to worry about creeps creeping on them and their family.

The APP is super simple because I am testing backend function. The scope is pretty large so I wanted to have a few different versions.

I've been learning as I go with a lot of this stuff but this is going to be built by someone Ive just been lucky enough to have some insight after researching this is absolutely going to change the face society as we know it.
I've spent quite a bit of time building out versions and I definitely need some help.

I should probably use simpler language when on-boarding people into the project.. The ability to prove actual work is pretty transformative. Also I'm not 100% if building for a Cellular platform or the Next gen of Orange Pi would be Ideal. Now I'm building for GraphineOS on a Motorola G.
It is a large project 4 sure.

AI is definitely letting people work inside domains they are not familiar with(often not qualified in)and People do sound ridiculous when they are talking about topics they are not familiar with but I've done the foundational work and its very apparent this is going to develop organically.

I think we have a unique opportunity, The AI industry is over looking like 95% of use cases because they still see Local AI as hobbyist when it's already past the point of being usable for something like this. The main thing is trust. Once you can trust that your data isn't being exploited using AI then tools for health and work can be built and used with the tech we have now.

I've honestly been underselling the hell out of it because it is so paradigm shifting.
The concept has Developed over the course of a few years and once I realized the potential I spent 6 months Solo Sailing the East coast and developing the concepts that would makeup the Ideal platform. Now I just need a bit of help bringing these concepts to reality. I'm a researcher by trade and I'm good at it. I started learning machine learning for 3D printing in 2019 then during the Covid lockdown I spent some time taking Harvard and Stanford classes to understand the foundations of the tech. If we can build this than we will be building the foundation of the next 100 years of human society.

Ok i kinda see what you are getting at. Almost like a pokemon GO style app that runs locally on a home computer or server, and connected to multiple devices that an organization uses to complete tasks and manage employees/members. Its kinda like what we had on our PDA's when I worked at sams club for the pickup team. I can see something like that working but the two problems I see are both stem from if its open or closed source.

• If its open source then the data can easily be spoofed and members that are nefarious can claim to have completed tasks or arrived at GPS coordinated they never went to, ect aka exploitation
• If its closed source then its harder to work on and less appealing to the local AI model community who tend to work better together as a team on public github repos. Plus you'd need hardware attestation or peer verification to make sure that the field devices dont get spoofed or modified either.

For a start id show some footage maybe in a youtube video of the actual program/code running with some field devices and not just a pre-configured demo, but an actual working senario that you can later refine. I think if you can have that to start, you'll get more traction, but until that happens I dont think you can start asking experts for help in something thats purely theoretical, and only has some UI screenshots, as anyone with access to chatgpt can generate those.

Wait ROMBO, NOTHING you read in the chats before your comment had ANYTHING TO DO WITH MY PROJECT... I was just talking with that guy about the Starnet UI. I updated the Org Card a bit ago. The actual onboarding Docs are not public but since you are part of the group I will add them in a private bucket.

ngl thats pretty funny

That's the great thing about adding organizations. One of the user's profile points could be their shared associations and organizations that trust or vouch for them.
Every use case is separate and every organization can add their own terms to a quest or task. It's primary use as a day to day guide would for sure have certain hacks and an evolving environment but it wouldn't really be practical for someone to spend a ton of time spoofing their data for a small task. It is meant to reinforce small communities so the trust system I am developing for transactions should have a trust rating for every interaction..(Should definitely do more development on this.)
It creates a pretty unique environment for security all together.

I was thinking that the developers would have a council, so if one of the users Organizations couldn't resolve a dispute it could be elivated to the Council of 9. or something.
But Look into the Android field assault kit. Something like that with intelligent tools and an automated support system with the Prover system.. transformative..

But Pokemon go but for real life, "You've just achieved Level 27 stoic, this would be a good thing but you are supposed to be selling Redbull"
I was developing skill books for the users title or expertise. The collective core should work like a craigslist but for tools environments and the uploadings could give the up-loaders "Kleos" points or Renown. Kleos points are a basic point system its one of the only things that gives the user an edge in his environment so the user user 10 Kleos points to skip ahead in line. They can only be awarded by organizations or some kind of community service.

It runs on the field device
A user would have a prep suite at home..
that's the primary structure. a field device and a supporting home environment. there is a Craigslist like collective core where developed assets could live.
An organization could also have a prep suite for the organization and it's members. But the foundation is the ATAK like field device then a safe storage and prep device. each org is unique with their storage and prep.
but the Data system isn't editable in the field. it can only be read so the profile is only uploaded by the prep suite when it offloads to the prep suite, it then reviews and builds out the lightweight packages for the next day. Like a lie detector or a meeting package with the persons profile talking points etc. for a meeting with your kids teacher.

It can use environmental data collected to do a ton of stuff.
If you work on cars it can diagnose engine noise
Or if you walk through a certain forest and there happens to be a survey for endangered birds or something you can review your own data for the bird and give a yes or no answer to the survey. or if it hears one you can sell that or use multiple days of sensor data of something to train a new model.
Also the Data marketplace is really the only way we can create a good Super intelligence and the users will get to sensor their own data. But when the user controls their own data it changes a lot. Like a real credit system based on who you actually are. Or being able to control insurance for exactly what you need it for.
There are a million different ways for it to make society better but the device itself will give people a feeling of Equity and freedom. Something that we need right now and freedom is contagious so once its out there it will fly off the shelves of whoever is selling it.

You could start an AI fight club guild and post the specs to the Core..

ORG CARD UPDATED AND SUPPORTING DOCS IN THE ONBOARDING DATASET

If you are intersted in telling how terrible it is and testing just let me know and I can finish the bucket for the local prep suite

I read through everything you wrote in this post, and as someone who has been in the AI field for a number of years now, having used the majority of Open and Closed source models extensively, my honest opinion about this project is that it sound like a slop machine.

This is everything and its nothing. You use fancy language to describe things that have been done 100 times before, and it contributes nothing to the field of AI. I recommend getting your mind out of the AI hype, and actually thinking of practical use cases for AI models, as well as building architecture around those models for systems that can be used in everyday life.

Rather than building the next swarm of Jarvis agents that every other tik tok star has already built and been made fun of by people who actually want to find productive uses for AI.

Like dont get me wrong, i dont want to be a buzzkill, and if you have something you want to create that is genuinely unique and helps solve a problem that has not been solved yet, Im all for it and Id support you in it. But this just seems like everything other project Ive already seen thats only driven by non-sensical buzz/hype.

@rombodawg Much respect for saying what we're all were thinking.

I know It's easier to dismiss a new Idea than to take the time to try and understand it.

It's has a ton of opportunity and I see that it's a red flag but the opportunity and large scope developed after the core concepts
It's pretty basic so I'll explain it like I would to a 3rd grader. I can't blame you guys for not taking it seriously after all we've seen a lot of out of domain people not willing to put in the developmental work but It's just bad practice to dismiss something simply because you can't understand it.

The Field device (Say a phone) has an ATAK like mapping and tool software.
The preparation ENV simply prepares research and assets for the devices..

It's not that deep. there are things though about the build that make it truely paradigm shifting. so instead of sleeping at a bus stop DaftCode you could show someone that you are not a psycho and they would offer you work for room and board.. It would actually give you equity and self worth, things you are sorely lacking.

Also I don't know what The first commenter was talking about but he was excited about having an AI assistant so we chatted about a seperate application of Starnet.

If you can't read don't respond.

The Actual programs I can share, I also have some basic Grok built html examples of the UIs I made while working on the themes of the UI if you don't want to actually run some probably sloppy software. I think its design is pretty considerate of security concerns as the personal level adds a unique layer to the application but won't know the attack vectors for sure until field testing. I appreciate your input and consideration.

Ok i kinda see what you are getting at. Almost like a pokemon GO style app that runs locally on a home computer or server, and connected to multiple devices that an organization uses to complete tasks and manage employees/members. Its kinda like what we had on our PDA's when I worked at sams club for the pickup team. I can see something like that working but the two problems I see are both stem from if its open or closed source.

• If its open source then the data can easily be spoofed and members that are nefarious can claim to have completed tasks or arrived at GPS coordinated they never went to, ect aka exploitation
• If its closed source then its harder to work on and less appealing to the local AI model community who tend to work better together as a team on public github repos. Plus you'd need hardware attestation or peer verification to make sure that the field devices dont get spoofed or modified either.

For a start id show some footage maybe in a youtube video of the actual program/code running with some field devices and not just a pre-configured demo, but an actual working senario that you can later refine. I think if you can have that to start, you'll get more traction, but until that happens I dont think you can start asking experts for help in something thats purely theoretical, and only has some UI screenshots, as anyone with access to chatgpt can generate those.

I'll simplify the Org Card this weekend. Either way I appreciate the feedback.

"A Data Secure society(Where private citizens will use secure data to employ systems they can trust within their environment and with what actually matters) is an inevitability. It is not an "IF" it is a "When". The current structure of companies holding important data simply will not work in the next decade."

That made me laugh IRL.

For that, I'll give you one, very good advice, that has served me in the field of software development for the last 30 years: "If you cannot communicate your idea in an elevator pitch that last less than 30 seconds when spoken out-loud, explaining a concrete use-case that's not already being handled by something else, you do not have an idea that's worth anything."

Your readme, ain't it. It's quite the polar opposite of it. I know it's written by hand, given all the typos, but the noise to signal ratio is very LLM-like, if that makes sense.

Yea I agree, and it's not that anyone is being dismissive of your idea. We want you to know what idea you are trying to convey before you start a project. Work out the actual steps of creating your product and it's software, plan what it's gonna cost, what you'll need to carry it out ect. Have a game plan, not just a broad summary for something that doesn't exist yet.