I sometimes keep 4-5 active sessions going plus a couple waiting in the queue. Can't say it's always convenient. I don't know about others but I sometimes forget which session is doing what and end up getting confused.
Hi, I’m the creator of AI DevKit. Problem that you are facing is one of the thing made me build this.
Once you have 4–5 active agent sessions, the bottleneck becomes remembering who is doing what and jumping between them.
ai-devkit gives each agent a name, lets you see and navigate active sessions from one place, agents can communicate with each other, search sessions that you already completed. I also added remote control so I can check or redirect sessions without constantly switching terminals.
If you have time or chance to try it out, I would love to know what you think.
Have you ever tried any of those ADEs / Agent orchestrators? I have considered downloading a few, tried Kepler for a bit, but I am unsure if I am maybe trying to search for a tool I don't actually need.
The article is not bad, but a little outdated. Syslog is a database that will not go anywhere, because a lot of equipment and software still send logs this way. But in 2026, building the entire infrastructure around it is already strange.
Honestly, I wouldn't call it plagiarism. You're using the prompt for personal purposes, to better understand the material. If you were copying his text and passing it off as your own, then yeah, that would be not cool. But using his approach with interactive visualizations as a source of inspiration for your own learning materials is fine.
If you ever decide to put it out in the open, just mention that you took inspiration from his style and include a link to his work.
Interesting breakdown of a hypothetical attack. The complexity of modern inference engines and the rush to develop them do create some attack surface. But overall it reads more like a "what if" thought experiment.
Splitting the GPU and parser is technically doable, but in practice it's trickier, large models run on clusters where the boundaries between components get blurry so defending against this kind of thing would probably require some serious rethinking of the whole architecture I think.
>But overall it reads more like a "what if" thought experiment.
I remember the days when people said opening files like images was safe because it's not an executable. Then people got clever and started exploiting said image libraries with things like numerical overflows.
So, it's really important to ask these questions in a general security sense and think of mitigations before someone finds it's possible and crashes most of the internet.
Solid interview but honestly a bit boilerplate in parts. The bit about "program death" and Naur's theory is gold though, even worth re-reading every few years.
His take on AI feels a bit cautious though. I think he's underestimating how fast tactical tornadoes are gonna become the default just because managers love seeing shit get done fast, even if it's messy...
It's funny how we're all chasing power, and yet here's a 60-cent chip emulating a Mac that people actually got work done on. Not hating on modern devices, they're awesome and all. But it's still interesting.
Also, kinda want to order a RP2350 board now too, but I know I'll probably end up having zero time to do anything with it and it'll just sit in a drawer somewhere...
