AI usage for Elten

Ok, please hear me out before shouting "AI slop!".
Unfortunately, we all know that Elten is horribly slow. If the open sourced code was up to date and the server wasn't missing, I'd already do it myself, but since it isn't, here my suggestion.
Would it be so bad running Copilot, Claude or ChatGPT over this and see what it can improve in terms of readability, maintenance effort and especially speed? I personally think that the coding abilities of Claude are underrated and overhated. As long as one knows the basics about how to prompt, wonders can be achieved.
Elten, however, is a huge platform, so of course each edit would ned to undergo careful review or weeks of beta testing, however, I am quite positive that eventually, Claude or another coding assistant could improve both resource usage, fix known bugs, the code disaster and - of course - the speed.
Maybe, with enough prompting effort, a migration to true Rubi or another language can even be completed within one or two weeks of Claude code on the pro plan.
See me as AI simp if you want, but I personally just see a huge potential especially for such simple yet annoying tasks such as lag reduction, but also for way more complex things.

I actually agree, if used correctly, AI can really help in this scenario.

I couldn't agree more! In fact, I am trying to do exactly this right now!
I’ve been working today to prototype new features for the desktop client. In just a matter of minutes, I successfully coded a native AI Assistant (AITutor) using the OpenRouter API. I injected it directly into MainMenu.rb and got it working perfectly with Elten's native input_text and TTS alert functions. The potential for AI to improve this platform—whether for refactoring laggy code or adding smart features—is absolutely massive.
However, I hit the exact same brick wall you mentioned. When I compiled my .edb file from the GitHub repository to test the integration on my PC, it downgraded my client. Because the public code is so outdated, the native network agents (like read_url) timeout or crash, making it impossible to fully test external API calls natively.
We don't even need Dawid to do all the heavy lifting himself. If he simply updates the GitHub repository to match the current live client, we could submit fully tested, working Pull Requests for lag reduction, bug fixes, and awesome new features within days.
I have the code ready. We just need the up-to-date source code to make it happen. Let's hope Dawid sees this and updates the repo!

I use AI in coding a lot, and when you use it correctly, it can be so great.

Exactly, I use it too.