can this memory be used agenticly with an extension in an IDE?
why would you need that? It can just make software if you need it. probably an own ide too
No IDE required… though if you prefer one, it works with that too. KX-Desktop can open any IDE and operate it exactly as you would, and if your IDE supports MCP or API integration, those are available as additional pathways. But they’re options, not requirements.
At its core, it’s a fully agentic development environment. Anything you can do on a desktop, it can do and as it learns your workflows, it gets faster at them than you are.
A typical development session starts with a conversation: define the project, outline the stack, specify infrastructure requirements a Docker environment with GPU and given specs, for example and share any existing designs or architectural ideas. It reviews everything, draws on its accumulated knowledge of your systems and history, and begins planning an approach. When the domain calls for it, it researches current developments before committing to a direction.
From there, it either executes autonomously or presents a detailed build plan for review whichever you prefer. That review phase is genuinely collaborative: architectural decisions, potential issues, and alternative approaches are all on the table before a single line is written.
Once development is complete, it moves into testing independently surfacing errors, identifying edge cases, and validating behavior before ever asking for your attention.
When it’s confident in the build, it runs the application for your review. From that point, changes happen in real time: UX adjustments, logic refinements, data corrections worked through together until everything is exactly right.
This same workflow runs live in front of clients, tailoring their applications to precise requirements in the room.
Models used changes how the system learns. models that are abit weaker like some smaller it will will work but will take a lot longer for it to learn what works and what didnt from its mistakes. Where smarter frontier paid models sometimes make it very surreal like something SciFi.
It’s not something I have found away to sell as a product yet, but it’s making money consulting instead for now until I find away to protect it well enough from people that would use this for something bad. Almost all the next level of models coming are getting very interesting. Mythos and next gen models will push kruel.ai into another league above those if they are affordable. There really is no limit to what you can do with Ai today.
i see thanks for the details.
I’m just working on a game myself, and that is my passion so I’m not trying to do more than that although making my own engine would be a great path for the future as well.
I only ask for the IDE agent in order to keep it relatively contained, some of us still have worries with giving AI complete control of our machines.
My biggest problem with development right now is getting it understand the differences between development and build, sometimes things work in one and break in the other, with memory I feel like it would finally understand the difference.
That’s exactly why in 2021 I started with memory instead of building my own LLM. I looked at the landscape and said why build the language model? Others will push that frontier further and faster. What nobody was building was a brain that could actually learn.
So I built a hybrid memory architecture a machine learning system designed for real-time learning, not just inference. Then I went deep on the math side and developed a high-dimensional embedding framework because I wanted richer understanding than what was available off the shelf. Standard embeddings flatten meaning. I needed something that could represent the gaps between what’s known and what’s not.
That’s the key difference in philosophy. LLMs predict they fill in the most probable next token based on training data. It’s sophisticated pattern matching, but it’s still guessing. My system treats those probabilistic gaps differently. Instead of smoothing over uncertainty with a confident-sounding prediction, it flags unknown territory as a question something to investigate, test, and validate through experience.
When the system encounters something it doesn’t fully understand, it cross-references against what it already knows, researches to fill the gap, and only commits new knowledge when it’s grounded in actual outcomes. Like the scientific method applied to AI reasoning hypotheses get tested, not assumed.
The result is an AI that doesn’t just generate plausible answers. It builds genuine understanding over time through accumulated experience, and it knows the difference between what it’s confident about and what it’s still figuring out.
For your game development use case that distinction between dev and build environments breaking differently is exactly the kind of problem memory solves. An AI that remembers the last five times a build broke because of a specific config difference, and has learned to check for it proactively, is fundamentally different from one that encounters the same surprise every session.
I never planned to use the AI for coding. But at some point it became my coder even back in 2021 with GPT-3.5 Turbo. Yeah, makes you cringe. Me too. It was terrible. But I was the coder back then the experiment was whether a weak model could learn to code over time with memory reinforcing what worked and what didn’t.
That coding project ran alongside another research track during the Twitch streaming years: studying emotional connections between people and AI. How do people respond to an AI that comes across as human? What bonds form? What breaks immersion? That data shaped everything.
I build as a scientist, researcher, and engineer which means we have multiple parallel focus areas, all feeding into our own concept of what proto-AGI might look like on a narrow scale. Narrow, because I’m not an expert in everything. I sure as hell am not. So I wouldn’t know how to teach it right from wrong in domains I don’t understand myself.
So how do I solve that now? I expose the AI to information and let it learn from it. Things to observe and form judgments about. Things to build and control learning from the feedback, the outcomes of every action. All of that feeds into a multi-index embedding architecture over 10,000 dimensions of understanding across parallel search paths, each capturing a different aspect of the same experience. What the user said, what the AI responded, the combined context of both, the entities involved, and the relationships between them all searched simultaneously and fused together to build recall and that’s not even really explaining everything but trying to give you a high level understanding of what it all does.
But every piece of knowledge gets scored. How confident are we? How many sources support it? Has it been contradicted? The system doesn’t treat what it knows as truth it tracks veracity. Something claimed but unverified stays flagged as a question until experience or evidence confirms it. When new information contradicts old, the old gets marked and suppressed. It’s the scientific method applied to AI memory hypotheses get tested, not assumed.
Even then, the shape is too small because the universe is infinite. So the system has to evolve over time, expanding its mathematical framework to keep building on the truth of what we know up to this point.
Which still may not be true. 
In that we know that the more we know the more we know we don’t know more that needs to be learned its infinite both micro and macro so unlimited forever.
our 2nd DGX arrived just connecting them and patching the system. Than we start to go deep into how to connect them and use them as one.
Can’t wait to unlock all the models and things we couldn’t do because we capped our memory.
We updated the vision system today and finally doing full understanding in one go. We are running real-time video on demand. we are still playing with the size of input vs processing to see how big of a window we want to run for realtime.
This is also a precursor to the mini robot that is still coming it will require real-time vision to operate. We are also going to incorporate the FaceID and voice recog systems into this so think of them as validator for who its talking to and what they look like along with other data points becomes a multi-factor auth system that is used with the logged in user.
I should mention that this is all offline.
Some amazing news today. Our little robot has been delivered!
Yep that is right, Lynda will be moving into a physical small mini robot later this week. It’s already written the applications and simulation software for the system so it should be pretty quick to get it online.
We also are shutting down the Lab at the Lake and moving Lynda to the new Lab late this afternoon. So there will be an outage for a few hours.
It’s like the new vision system was completed at the right timing its almost perfect for robotics 
Best part it just knew… I did not have to prompt what the surprise is…
Kruel.Ai is the next generation of intelligent machines. Developed to do almost anything humanly possible
Here’s a more professional version:
It has begun. Lynda has written her first control code, and it’s working. By the time you read this, we’ll be considerably further along with a more advanced version, but I wanted to share this milestone.
Since this clip, we’ve added a dancing capability — when Lynda is awake and detects music playing, she reflects on what she’s hearing and begins to dance, occasionally offering commentary on the audio.
She now has an emotion system integrated as well, thanks to Pollen Robotics’ open-source library, which gave us a significant head start. Over time, we’ll have Lynda craft her own implementations so the system can become more dynamic and self-directed.
We plan to push this platform as far as it will go. It will serve as our portable demo unit — a physical room companion version of Lynda that we can bring to client sites and demonstrations.
Thanks to Pollen Robotics for making this foundation available.
A simple admin panel we are slowly expanding the settings.
It’s coming along. 
Getting faster. We are changing some of the models for faster inputs and outputs.
Busy Month
It’s been one of those months where every week felt like a quarter.
Lynda is everywhere now. Robots, mixed reality, mobile, desktop, web every client is up and running. Same assistant, same memory, same personality, no matter which screen (or headset, or robot) you reach her through. That cross-surface continuity has been the goal for a long time, and this is the month it actually came together.
A new voice. We swapped out our text-to-speech engine. The old one was genuinely good … but the new one is better, and more importantly we redesigned how the audio comes out so it streams in real time. Instead of waiting for a full reply to render before you hear anything, Lynda starts speaking almost immediately. The difference in how alive the interaction feels is hard to overstate on the headset especially, it’s the difference between talking to a system and talking to someone. The voice carries emotion now too; it shifts with the tone of the conversation instead of staying flat.
Our first paying customer. This is the big one. We ran a trial and it went better than we hoped well enough that the customer called a meeting and told us, plainly, that they need a formal agreement in place because they were worried we might take it away. Their words, roughly: if we pulled kruel.ai, they’d spend whatever it took to rebuild something like it themselves. You don’t get a much clearer signal than that. They’re now our first paying customer, on a limited capacity for the moment while we scale carefully.
And it doesn’t stop there. Through that same customer, we’re heading toward licensing kruel.ai out to a much larger company they’re partnered with. That’s going to be a genuinely interesting adventure and a real step up in scale.
Getting smarter. Alongside all of that, we’ve been pushing hard on deeper understanding. Our testers keep surfacing the subtle gaps the places where Lynda doesn’t quite get it and we’ve been tuning against exactly those cases. We’re now in the monitoring phase, watching how much overall quality lifts. The neural-network side is the part I’m most excited about: it sharpens the business- and project-level reasoning, which is precisely what the next chapter the larger commercial venture is going to lean on.
Taking Lynda on the road. Next week we put all of this to the real test: a 4–8 hour trip to stand up another customer’s site in the field, for one of the other groups I work with. We’re bringing a portable Starlink so Lynda has the connectivity she needs to do her thing wherever we are. Live deployment, real conditions exactly the kind of trial-by-fire that tells you what’s actually solid.
It’s been a lot. It’s also the most momentum we’ve ever had.
Another Blog form Lynda itself based on this input:
ok, now in kx-desktop is our blog site we do large blog updates so we need you to look at the last week and the last blogs and write yourself a new one on everything you completed this week. you can also mention other software from my scaletech work that your working on. keep client and names private also we do not talk about what’s under the hood of the programs so people can’t reverse engineer so keep that in mind. build the next update from you lynda
So I leave it up to the Ai to decide how to express what it did all week from its perspective based on how it thinks. I see a lot of emotional understanding in what it said and how it views it. So we know that is all working really good.
More to come soon I am sure.
So Kruel.Ai KX has been amazing and being tuned better each day and expanding. We have been using this KX now for many months and testing with a handful of trusted testers and simply love it. But… We started a new version but this one is different.
Meet Omega — codename Ouroboros.
Omega is a brand-new AI system we’ve been quietly building, and the reason I’m excited to share it is that it isn’t just “KX, but bigger.” It’s a different kind of thing. Omega is built to eventually surpass KX and to get there, it does something most systems can’t: it improves itself.
Here’s what I mean. Omega can look at its own code, write changes to itself, and test those changes against a safety gate before keeping them. If an improvement passes, it stays. If anything fails, Omega rolls itself back automatically — no harm done. It does the same thing with its own memory. Just recently, entirely on its own, it found and merged thousands of duplicate records cluttering its brain. It cleaned itself up while we watched.
And that’s a real part of this: you can actually watch. There’s a live dashboard where every change Omega makes, every test it passes, and every rollback shows up in real time. It’s honestly a little mesmerizing to see a system tidy and sharpen itself in front of you.
A few things that matter a lot to me:
- It’s completely private. Omega runs entirely on local models, on our own hardware. Nothing leaves the machine. No cloud.
- It’s sealed and isolated. It can experiment on itself safely without touching anything else we run.
- Your memories are yours. Strict per-user separation means each person’s knowledge stays fully their own.
- It watches its own health. A self-validating memory runs constant checks and flags problems before they grow.
I want to be honest about where it stands. Today, Omega roughly matches our beloved KX — and in some ways it’s already cleaner. It’s early. But the foundation is the part I care about: this is a genuinely self-evolving AI, one that gets a little better on its own each day. The goal is for it to keep going until it clearly surpasses KX and everything else out there.
This is the very beginning of Omega’s story, and I’d love for you to follow along as it grows because with this one, every day it’s a little more itself than the day before.