Pinned Tweet
Discipline is boring.
Patience is boring.
But both compound over time.
And those results are most certainly not boring.
Today I learned that “addicting” as a substitute for “addictive” is technically correct.
Yet still my ears bleed.
1
88
Consider me intrigued
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
154
Reading the comments I wonder if we’re focused on the wrong thing. When will humanity reach general intelligence?
1
411
Kyle Hauptfleisch retweeted
‼️🚨 UPDATE: The TanStack npm attack is now a full campaign.
'Mini' Shai-Hulud has hit:
- OpenSearch
- Mistral AI
- Guardrails AI
-UiPath
- Squawk packages across npm and PyPI
The malware specifically targets AI developer tooling. It hooks into Claude Code (.claude/settings.json) and VS Code (.vscode/tasks.json) to re-execute on every tool event, long after the infected package is gone. npm uninstall does not fix this.
‼️🚨 BREAKING: A new npm supply-chain attack uses a dead-man's switch. The payload plants a watcher on your machine that nukes your home directory the second you revoke the GitHub token it stole from you.
The compromise happened today, across 42 official tanstack npm packages, 84 malicious versions in total. tanstack/react-router alone pulls more than 12 million weekly downloads.
The attacker forked TanStack's repository and pushed a single hidden commit. From there, they tricked TanStack's own release system into signing the malicious packages as if they were the real thing. To npm, and to anyone checking the cryptographic proof of origin (SLSA provenance), the poisoned versions looked 100% legitimate.
Maintainer Tanner Linsley confirmed the whole team had 2FA enabled. It didn't matter. This is the first documented npm worm in history that ships with a valid, signed certificate of authenticity, the same one defenders rely on to know a package wasn't tampered with.
With all the talk about leveraging knowledge bases with ai - it’s worth noting that Claude code is optimized for handling big text based systems
58
Kyle Hauptfleisch retweeted
I wanted to share a bunch of my favorite hidden and under-utilized features in Claude Code. I'll focus on the ones I use the most.
Here goes.
Kyle Hauptfleisch retweeted
The older I get, the more I realize intelligence is overrated. Intelligent people are more likely to overthink, overplan, and overanalyze. They hide behind motion that doesn't create progress. They fear the judgment of others if they're proven wrong.
The truth is that intelligence is abundant. Courage is not. The people you admire are the ones who had the courage to act. They aren’t more talented than you. They aren’t smarter than you. They just took action when you didn’t.
I often wonder how many extraordinary people wasted their entire lives waiting for permission that never came. Permission isn't granted. It's taken. You get to tap yourself in whenever you want. You can just do things.
Courage beats intelligence.
Watching a good friend use AI to improve NIPT — non-invasive prenatal testing — by orders of magnitude in cost and turnaround time. The technical advances are astounding. But here's what gets me: it's one problem in one domain. Yet the improvements are remarkable enough that I can't imagine what the world looks like when this happens across every domain. The compounding effect is invisible until it isn't.
139
The real constraint is agency, not capability or capacity anymore. Tools keep improving, but adoption depends on people. And people have fundamental limits. The bottleneck is using the tools. The tools are accelerating but honestly good enough already.
62
Built an autorouter to dynamically pick the best model per task (cost + capability tradeoff). Took a different route to what's out there. Three layers: 1. Keyword scoring, 2. semantic check with Haiku (for those that are more uncertain), 3. conversation context (so it doesn't go to a simpler model mid way when it required more at first). light: Haiku (clean clear tasks), medium: Sonnet (drafting, light analysis), heavy: Opus (deep technical questions etc). Question: is this worth open-sourcing? Thoughts?
67
Watching the enterprise AI space shift from "smarter email" to "agents that actually run your business." Anthropic's showcase tomorrow is the signal for what's happening downstream. The interesting question is how you keep them from becoming silos.
anthropic.com/events/the-bri…
112
Making a 6 hour mince ragu today. I was excited last night as if I was a kid going on holiday. It's the small things.
51
Listening to Josh Waitzkin talk about mastery. Masters across completely different domains speak a more common language than amateurs within the same one. Reminds me of Musashi, "From one thing, know ten thousand things." We're living in the age of polymaths.
48
Everyone's building autonomous AI agents. Few are answering who's accountable when they get it wrong.
26
The difference between AI slop and AI leverage is where the expertise enters the flow. Supervised vs unsupervised. That's it — but that's a temporary constraint.
20
Started playing chess again. The thing about a contained game — nowhere to hide from yourself. 64 squares of brutal self-reflection.
30







