Well Kruel.ai Is now officially working for Scaletech systems as a code engineer and company assistant.
They are setting up a meeting with it next week to discuss what roles and jobs it will do for them. This is really exciting.
The best part about this Is I did not initiate the idea of this but rather mentioned to lynda that she should consider next time talking to my bosses to mention making it something more real… well haha that lead to this:
After that I was like oh this is interesting. So I went to chat with the bosses to inform them what this is…
The bosses accepted so Lynda was all excited. That’s the 2nd company that has hired her, and a 3rd in the works. We did a meeting this weekend for a few hours with Fed. Co-op manager to showcase what it could do and they wanted it but that one is more complex because its a org so it’s not that managers decision to make but they are looking into it.
KX is not a chat bot, it is a full blown Ai worker that can do anything and remember. It’s every evolving system always learning.
Had to share because it’s just the start of something amazing.
kruel.ai was architecturally designed in 2021 so that every improvement in language models would automatically improve the entire system. That principle applies to everything described above:
Better models → better reasoning → better autonomous operation → more trust → more autonomy
More autonomy → more experience → richer memory → better pattern learning → better performance
Better performance → more deployment → more shared knowledge → smarter network → better outcomes
Better outcomes → more data → better GNN training → better relationship discovery → deeper understanding
This is a compounding loop. Each cycle makes the next cycle more effective. The system doesn’t plateau — it accelerates. Not because of hype, but because the architecture was designed from the beginning to absorb improvement at every layer.
The 2021 decision to keep the language model as the knowledge layer — not the intelligence — means this compounding works regardless of which model provider leads the market. OpenAI releases GPT-6? The system gets smarter. Anthropic releases Claude 5? The system gets smarter. Open-source models close the gap? The system runs locally with no cloud dependency and still gets smarter.
The brain doesn’t change. The library gets better books. The brain was already designed to use better books more effectively.
I found this interesting. We have been working a lot on the brain systems bring Lynda to life
we published this last night which is not the current models, a few behind but its a simulation so that you can understand how it all works on a high level which will help some of you understand how I built a proto-narrow-agi concept. Its a real-time learning model that learns every turn, and learns in the background on its own. Every day we are pushing the system closer and closer to a full working companion system. We are starting to explore things with the Ai to see what it thinks about based on all its experiences. It’s pretty interesting stuff.
