Liang is Sufficiently Financialized

Liang is Sufficiently Financialized

8,815 Twitter followers

Follow

founder of @getus3r, summoning @realdegencast, prev CTO @SynFuturesDefi, ex-pleb @binance @BITMAINtech On Farcaster: http://warpcast.com/liang

Posts

Liang is Sufficiently Financialized

OpenClaw has clearly overhyped itself. However, between the two extreme viewpoints (this is AGI vs. this is just xxx), the most interesting aspect of OpenClaw is that it elevates the agent's decision-making level (an agent that uses your Mac, not just an agent that writes code). A higher decision-making level means more flexibility, and the user interface naturally transitions from a code editor to a chat application. twitter.com/QiaoLiang3/status/...

Liang is Sufficiently Financialized

In AI coding, in 2.5 years we went through: - It can't even write more than 1 line of code at a time. - Okay it can write entire functions but can it do anything useful? - Sure it can fix some bugs and develop features but can it write entire projects? - Yeah it can write a C compiler in a week but it's not more efficient than GCC so it's pretty useless

In 18 months we went from - AI is bad at math - Okay but it’s only as smart as a high school kid - Sure it can win the top math competition but can it generate a new mathematical proof - Yeah but that proof was obvious if you looked for it… Next year it will be “Sure but it

Liang is Sufficiently Financialized

Recently, many people have successfully implemented a scenario in Alibaba's Qianwen app where AI can order milk tea, and many are eager to try it out, perhaps creating a new application scenario for AI. Isn't this the x402 scenario that the crypto has been longing for? But in Qianwen, you can confirm the address and make a one-click payment; no cryptocurrencies or stablecoins are needed throughout the process. 🤔 The assumption that "AI absolutely needs Crypto" needs further verification.

Liang is Sufficiently Financialized

Authorop's Research Ralph looped a 100,000-line C compiler: ``` #!/bin/bash while true; do COMMIT=$(git rev-parse --short=6 HEAD) LOGFILE="agent_logs/agent_${COMMIT}.log" claude --dangerously-skip-permissions \ -p "$(cat AGENT_PROMPT.md)" \ --model claude-opus-X-Y &> "$LOGFILE" done ```

可能软件白菜价的底线就是token成本了一个C Compiler大概20000刀 https://anthropic.com/engineering/building-c-compiler…

Liang is Sufficiently Financialized

Heads up: Anthropic is giving out $50 of free credit to many Claude Code users if you go to the usage dashboard and click "claim"

Liang is Sufficiently Financialized

The most painful lesson I've learnt chasing PMF 🥲 I spent 4 years mastering Zero-Knowledge cryptography. Then I tried to sell it. Nobody cared. I knew "solution looking for a problem" is wrong. Everyone knows it. But I thought MY case was different because it's not just some technology, it's the most awesome technology in the world. It wasn't different. So I dropped the solution and started doing discovery interviews with CISOs. What I learned: 1. No one wants to hear about encoding execution traces into polynomials. I had to invent a story about a cat in a room to explain how it works. 2. Cold outreach CISOs were brutally honest. Warm intro CISOs were polite. Brutal honesty wins. 3. LinkedIn is actually useful (I know, I'm surprised too) 4. The most popular CISO name is Michael. If you're a technical founder in love with your technology, the market doesn't care. Surprize 🍩

Liang is Sufficiently Financialized

Israelis military service intelligence unit is similar. produces lots of devs and founders

@CharlesDardaman

It's their coop program where the students coming out already have years of real experience working. This can also be done via internships through college, but that's all on the student vs waterloo where the majority of them come out with experience and you don't have to hunt as x.com/GergelyOrosz/s…

Liang is Sufficiently Financialized

i miss the old opus one-shot the code opus call the right tools opus go straight to prod opus

Liang is Sufficiently Financialized

A skill is a reasonable abstraction of a prompt. A skill is an implementation of automated context engineering. Skill technology is not complex; it doesn't have that "high-level" feel to it. From a paradigm perspective, skills are elegant, simple, yet very reasonable and powerful. This makes them anything but monotonous. Simple but not easy. twitter.com/QiaoLiang3/status/...

Liang is Sufficiently Financialized

Tech <3 MAGA: To avoid becoming an accomplice in the global and Chinese draining of the American middle class and hollowing out American manufacturing, we must shout MAGA.

一句话概括a16z这篇：为什么我们能募集到150亿美元？因为爱国为什么我们要募集150亿美元？因为爱国 x.com/a16z/status/20…

Loading..