Into JavaScript/ReScript • Co-founder, CTO https://xitter.cf/t.co/KVwpXg3uCe (YC W21)

London, UK
Joined June 2008
Karolis Narkevičius retweeted
we've gotten so obsessed with speed that taking time to think now feels like a bug. it's a feature.
11
4
89
3,253
Karolis Narkevičius retweeted
Time to start light prepping at home (water, food etc for a few weeks) in case of global cyber incident affecting supply chains
1
5
194
Karolis Narkevičius retweeted
Some stats on #php2shell. In the 10 hours before it was publicly announced, across the @Cloudflare network we saw thousands of attempts to exploit it. That jumped to tens of thousands in the 45 minutes post announcement, hundreds of thousands within 2 hours, and millions now.
290
3,272
672
22,656
2,264,041
Are the vibes a bit off with sol 6.1? Feels less intelligent, more verbose, less helpful? And opus 5.5 is a bit more precise, more detailed, but also too verbose. Struggling to enjoy either atm!
1
72
AI 10x-ing our productivity in the office, not! haha Astra is something else.. Try here: yard.kn8.lt/
10
143
Karolis Narkevičius retweeted
New post: going into our investigation of the HF attack (before Black Hat), I was very wrong about what basically happened. This incident was far more serious than I expected, and far more serious than previous documented misalignment incidents. planned-obsolescence.org/p/t…
87
411
131
2,417
947,665
Karolis Narkevičius retweeted
Some time ago I noticed that I almost entirely stopped going to new restaurants and drinking natural wine. It seems like it’s not just me. Recorded a little video essay about this.
1
1
8
919
Karolis Narkevičius retweeted
GPT-4 finished training four years ago today.
516
612
282
14,717
2,399,385
😱
Tomorrow will be my last day at Google after 27 years, and watching it grow from 25 people to 190,000+ has been an amazing journey. Below is a note I shared with many people internally at Google today. An excerpt is: It has been an absolute pleasure to work with you and to help build some of the most widely used and impactful products of all time. As a kid, I dreamed of helping build software that would be used by many people, and Google now has thirteen products used by more than a billion people (amazing!). Our work has had a tremendous impact in the world, and I have been lucky enough to collaborate and form friendships with many colleagues that I deeply admire, respect, and enjoy. It still brings me joy every time I see people out in the world using our products to find information, handle email, translate documents, watch videos, learn new things, navigate and understand the physical world, browse the web, use their phone, run large-scale computations on our infrastructure, ride in an autonomous vehicle, or perform complex tasks with the help of our AI systems. I hope you all share this sense of joy, because it is a shared accomplishment! Thank you to all of my colleagues at Google over many years! Now I'm excited to go start @DiscoLoopAI with my longtime friends and colleagues @Sanjay_Ghemawat, @OriolVinyalsML, and @quocleix. (Updated post: slightly redacted to not have some personal info)
2
141
Karolis Narkevičius retweeted
I feel like we are on one of those pendulum swings right now where people undo a decades worth of work only to realize in 5 years they over swung because they missed the actual problem. We see this all the time. I'm not going to argue against imperative handles in all cases. But I want you think about the implications. If you unwind these things you remove guarentees. Like why do developers like Rust or TS? Its because through constraints certain classifications of issues aren't possible. The solution solves itself structurally. Structural constraints also have a nice property when composed. You keep benefitting from them as you layer on complexity. Build more constrained systems. Like if you aren't batching synchronous updates you probably have no foundation to managing async race conditions structurally. We shouldn't confuse incidental simplification(removing a problem) with structural simplification. Sometimes legitimately we don't need to solve certain things. The solution should fit the problem. But more often than not in a well established space we are just delaying the inevitable.
Replying to @ryanflorence
Reactivity is overrated. Love this banger from Fable: > ... view updates when I say handle.update() — I have said it 281 times in this codebase and it has surprised me zero times. I do not miss guessing when React would re-render.
17
19
1
332
35,328
What do you use to give your agents a credit/debit card with limits etc?
1
60
Ah - just a virtual card in any of the (neo)banks. Easy.
52
Karolis Narkevičius retweeted
Claude Fable 5 debuts at #1 on DeepSWE. It outscores the previous best by 3% and sets a new state-of-the-art on our long-horizon coding benchmark.
108
84
55
1,617
476,066
Wild that it was so easy to switch to 5.5 in the last few weeks. Codex desktop app is also very neat. We’ll see what Opus 4.8 brings (next week?)
For complicated agent work, it's amazing how much GPT5.5 has improved. I found 5.2 to be very far behind Opus. Now using Opus 4.7 after 5.5 feels like a big step backwards. Gotta love this level of competion! Strong comeback for OpenAI.
3
3
132
Karolis Narkevičius retweeted
OK fine here is my thought. To a close approximation: - When code *is* the product, you are going to want expert human engineers carefully controlling the result to maximize reliability - When code is a means to an end, you won't Notice that this rule can be run in reverse: because humans control the result of products that are code, I theorize you'll see increased demand for "productization" of more and more infra. Basically a bifurcation of the software space: infrastructure and slop. And I don't mean slop in a disparaging way. There will be lots more quick software getting written by everyone. It won't ever be perfect, but it will get the job done. But in order for this to work well, it needs to sit on an absolutely solid foundation. More solid than the software community has produced before. This foundation will be very carefully managed by expert practitioners. It will have to be, because we will have 1000x more software running on it than we ever have before, created autonomously.
5
5
1
71
6,458
Karolis Narkevičius retweeted
I regret to inform you that Ask Jeeves is dead. The site closed yesterday. Web 1.0 lost another founder. Ask Jeeves: 3 June 1996 - 1 May 2026. Send no memes.
1,027
12,508
1,881
69,523
3,346,733
Anyone found where to migrate to? 🤔
yesterday I reported on GitHub employee concerns about reliability and leadership, and then hours later GitHub suffered a catastrophic outage 😬 theverge.com/news/918001/git…
1
49
Karolis Narkevičius retweeted
MiniMax M2.7 is available on Ollama's cloud, and is licensed for commercial usage. Use it with OpenClaw: ollama launch openclaw --model minimax-m2.7:cloud Coding agents, such as Claude: ollama launch claude --model minimax-m2.7:cloud Chat with the model: ollama run minimax-m2.7:cloud
We're delighted to announce that MiniMax M2.7 is now officially open source. With SOTA performance in SWE-Pro (56.22%) and Terminal Bench 2 (57.0%). You can find it on Hugging Face now. Enjoy!🤗 huggingface:huggingface.co/MiniMaxAI/Min… Blog: minimax.io/news/minimax-m27-… MiniMax API: platform.minimax.io/
Readers added context they thought people might want to know
The MiniMax M2.7 model weights are publicly available, but under a license prohibiting commercial use without authorization. This does not meet the Open Source Initiative's definition of open source, which requires allowing commercial use. License: huggingface.co/MiniMaxAI/Mini… OSI: opensource.org/osd
71
99
14
1,177
271,308