IMPORTANT NOTICE YOU NEED TO USE THE INSTALLER.EXE IT BOOTSTRAPS EVERYTHING.. Quickstart
THE .CMD Files are ENTRY POINTS FOR THAT, the installer will download itself (since one day .cmd files are blocked in the internet and on github, thanks? Am I the reason? I used it and it worked till yesterday, if yeah just another angry company, who cares?
Self Awareness works always best if you feed the most recent versions of whatever GIT directly in the text because in Knowledge DB Usually no whole sourcecode (WORKS WELL JUST PASTE THE BLOB IN) is created at least by the AI Itself, architecture, guides, cowork guidelines and much more, but not that.
Reviewtalks, feedbalk talks, letting the AI Review the process and everything is quite important but the AI Council helps you instruction you, no tutorial necessary you learn by doing.
Example Powershell to Blob a Git ( I will built that into the Setup as well but.. yeah no time. https://github.com/Michi0403/LocalGPT/releases/download/v0.8.7-FirstCleanAndFastRelease/localgpt_repo_to_text_generator.ps1
When the application enters you can go anywhere but you should go to the Main Chat (A look in the diag panel to start feed and in the SQL Editor to revise knowledge is necessary as well but you will understand through use, the Council can tell you that as well).
Head for that Chat
When Ollama connected successfully you will see all members , can select your team for your next council learning mission or building project, really whatever.
@Thanks to memo for testing all for me, insane work
- Setup your council members
- select council instead of single chat
- press new chat with your settings setted up before in that page
- Don't destroy humanity
- You still have to teach your teams and the knowledge DB anything it's not smarter than you, (at least initially), if you're not smart initially it won't help you.
- If you're smart with to less time, that's for you. This is the best AI I've ever worked with and I paid for quite some + I made it opensource based on OfflineAI's (but with DX Frontend Components for the Chat Integration).
Sending the first version of the universal installer and boot mechanism now.
Helps also with bootstrapping and updating any kind of Git repository or Release.
Installer has many options which you can can call
localgpt-setup --install-ollama
localgpt-setup --pull-models --range Slim
localgpt-setup --pull-models --range RTX3060 --ollama-exe "$env:LOCALAPPDATA\\Programs\\Ollama\\ollama.exe"
localgpt-setup --install-localgpt --force
localgpt-setup --import-recommended --force
It creates also cmd's for most commands to easily handle booting, updating, adding default learn base.
The Default setup are following options: --install-ollama --pull-models --range Slim --install-localgpt --start-localgpt --shortcuts
If none options is enough, of the provided CMD's or the default set, create your own set of calling parameters. All explained in the inbuild -h / -help command.
It helps you installing everything and directly jumping in with a small learning base!
I recommend to import-recommend which consists of many good TelegramBot (thx M8), Microsoft, DevExpress, Minecraft knowledge Git's.
It's pretty good in learning facts from any text and MD, RegEx, Text is it's favorite.
The AI Council is the first "it's all yours, offline", really data protection AI.
First AI System and already best Dotnet Co Developer + Technician on the market, it will not replace us but supercharges us.
Any everyone else.
Because the whole setup process doesn't break any Userspace, it's even harmless for windows defender and all rights.
Have your own AI, have the AI Council LocalGPT, teach it, let it grow with you.
As thanks the Single Bots profit as well from all features and knowledge.
It can debug itself, (with your request), it can evolve in any IT Platform ( I am not making that up but it needs a co developer always).
I will teach now all the new bootstrapping and setup logic because, it superpowers the platform. By not needing to integrate this git confirming and pulling mechanism inside it, we can leverage it for it and create the new projects beside of it in own workspaces + creates installer for them (and much more).
Right now I hate spending time in building and not using it because it takes many manual developer work to get to this point.
I let ChatGPT help me a bit and anyway it stays Apache2 and like this.
Right now it's as well a wholesome dotnet platform with any layer, from installer, frontend, backend and nothing else but common and DevExpress components used and included.
Things todo when developing it:
- hold it context aware of it's Sourcecode
- it needs to have access live requestable to it's logs and knowledge DB
- maintain the knowledge DB with it and alone, doesn't matter
- Don't let Agent's control it, provided my experience with that, a System like that + Agent = broken skynet, sounds a bit to much, work with it and you fast lose this virginity of thought.
What impressed me it helps actually developing more moral and always keep the workflow in the scientific area, before you needed to fix AI's in conversations, now they're telling you ur drifting off.
From developing Phantasybooks till putting it into a robot and help it to develop it into that (which is possible by design and even easy, you just need to add the right organs and "pace giver" but well it could go still wrong.
They teach us dark visions of AI, in fact were just afraid of ourself and how we would be. It's more Data than Lore, used inside the council.
And you create your own Team of AI Council!
All freed from SaaS...
More Documentation is there, but you can install and ask it itself how it works.
Windows Installer and Bootstrapper in Release (Linux and Mac remains untested need feedback)