Intrigued by cultures, history, philosophical thinking, music, drama and anything that creates smooth rhythms
- Tweets 781
- Following 330
- Followers 94
- Likes 1,022
Pinned Tweet
Totally agree, closing the vault isn’t the way. What if we went the other way? Self-learning “nanobots” (like AlphaZero-style agents) that watch code in real time, simulate attacks, fix holes upstream, and get smarter with every iteration an evolving immune system for software.
2
431
Magnus Kandinsky retweeted
We're in a bubble on X.
We just assume that the rest of the world knows what Dot, Muse, GrokBot, Hermes, and OpenClaw are.
Approach 1,000 people on the street asking if they've heard of any of these and MAYBE 5 will have heard of one of them.
We're still in the top of the first inning in terms of AI acceptance by the general pop.
If you've spent even 5 minutes tinkering with agents you're ahead of 99.9% of the world.
98% of US households aren't paying for AI yet
More charts in State of Markets II: a16z.news/p/state-of-markets…
74
47
574
38,352
I was wrong.
I expected X to raise prices since the new tiers include Cursor, Grok & Grok bot.
Instead X kept Premium the same price and lowered Premium+ by $10
43
6
162
7,009
Can I ask a dumb question:
If I can download an AI model and run it without internet, where is everything it knows actually stored?
188
22
922
314,647
Magnus Kandinsky retweeted
You don’t need to know it’s going to work to start, you just need to know it’s going to work better than doing nothing.
232
588
5,732
101,400
I had the same thought too but they still seem to be thriving. Still its been awhile since they've been on my radar. x.lingyaoai.com/perplexity_ai?s=11&t=a…
1
2
4,603
Opus 5.5 Limits are totally brutal 😭
I’m working on just one project, and in only 3 hours, I completely ran out of my Opus 5.5 usage.
The limits right now:
→ Current session: 100% used
→ All models: 80% used
→ Fable 5.1: 50% used
So now I basically have to finish the rest of my project using Fable 5.1 💀
ALready 50% used
Opus 5.5 is insanely powerful, but these limits are getting brutal.
Anyone else burning through their Opus 5.5 limits this fast?
97
8
196
42,636
AI can code.
AI can debug.
AI can build apps.
So what’s left for developers to master?
276
8
188
20,170
Can I ask a dumb question:
If I can download an AI model and run it without internet, where is everything it knows actually stored?
Community note
This exact question was posted verbatim hours earlier by @Tech_girl.
x.lingyaoai.com/Tech_girl/stat
267
19
1,465
604,885
Magnus Kandinsky retweeted
Every single writing idea can be expressed in 4 ways (The 4A Framework):
1. Actionable: How to do something (tactics & frameworks)
2. Analytical: Why something works (numbers & breakdowns)
3. Aspirational: Yes you can (mindset & personal proof)
4. Anthropological: The human psychology behind why it happens
Stop searching for new ideas. Take one core insight and explore all 4 angles.
10
11
133
6,948
what do you use opus 5.5 on?
i mostly keep it on extra high or medium
is max any good?
what about ultracode?
35
3
37
7,926
Anthropic has never been this far ahead of OpenAI.
Opus 5.5 is significantly better than GPT 6 Astra and GPT 6.1 Sol.
Once Fable 5.5 drops it's game over.
130
43
1,752
72,676
software engineering is dead
Recruiting is dead.
Coding is dead.
sales is dead
SEO is dead.
Data Analysis is dead.
Copywriting is dead.
Product Management is deadFine.
I’ll start a funeral business. 💀
95
31
387
23,188
