here's how i shipped 2,500 PRs last month to production
this was originally supposed to be for Cursor Compile in London. i couldn't make it since i was livestreaming for Grok @Bot Galaxy so i'm making it available for free here on X! watch it on 2x speed, i talk slowly
Jaka Hudoklin retweeted
My finest achievement. This is Windows Terminal using ghostty as an engine rendering a RDP client window using kitty protocol of chromium running in an agent sandbox in a container in wsl2. Thank you. I'll be here all week.
Jaka Hudoklin retweeted
Replying to @rohanpaul_ai
Disagree. He's failing to account for the far lower electricity costs in China due to China's vastly better power grid. He's also assuming Huawei just sits around and does nothing with its own microchip architecture.
Jaka Hudoklin retweeted
The excitement around Kimi K3 seems to support my "intelligence saturation" hypothesis.
TLDR - most people only need so much intelligence (capability, context, benchmarks, tool use, reasoning, etc) from models. Beyond that extra capability is wasted on them.
At that point, intelligence becomes a commodity. Sort of like how all CPUs are 64 bit today, and no one really thinks about it. At least on the consumer side.
In tech infrastructure, we break this down into two dimensions: lateral scaling and vertical scaling. Lateral scaling is adding more parallel processing or nodes that are cheap. Vertical scaling is making one thing bigger and more impressive.
At this point, we're nearing the point where open source models satisfy like 90% or more of all demand. Then it's just a matter of "who's GPU is it running on?" and that becomes the key market niche.
By the end of the year, and certainly within 12 calendar months, you'll have a library of open weight models to pick from that run at thousands of tokens per second, which will do the vast majority of the work you want.
there's a lot of excitement around Kimi K3 so we are making it available immediately to OpenCode Go users
we have not yet been able to negotiate a discount yet so it will currently consume your limits at a higher rate
hopefully will figure this out soon
Jaka Hudoklin retweeted
Meet kbd-1.0-codex-micro, built with @work_louder.
Map the buttons and joystick to your workflow, and keep your pinned chats in view.
Get yours before stock returns 410.
This is pretty concerning. You could still do this at the API level to some degree, but they seemingly just blatantly put it right into the code?
This is why open harnesses and agents are a much better option, among countless other reasons. You can inspect the code, observe the traces, and disable or modify anything you want for your own uses.
If you haven't yet - Hermes Agent is a world class coding agent. I'd recommend giving it a try.
‼️ BREAKING: Anthropic has embedded hidden spyware-like code in Claude Code that covertly targets Chinese users. It then sends information regarding every user by injecting it into their prompt message.
Claude Code is sending info like timezone, proxy and possible AI Lab connections into the system prompt in ways Chinese users can't notice.
A coding agent with repo and command permissions should not silently hide routing metadata inside prompts. This is a serious breach of user trust.
Jaka Hudoklin retweeted
New Moore’s law:
The cost to buy a chip doubles approximately every 2 years, while the number of people who can afford them drops by 50%
Jaka Hudoklin retweeted
what kind of attack is it when claude trains on all of humanity's outputs?
Jaka Hudoklin retweeted
Webdevs: PLEASE STOP setting your website language based on the user's location.
































