Bill Sun

I am really impressed by the execution efficiency of @zakfolkman, Cofounder @worldlibertyfi, which made me very bullish on USD1 becoming major stable coin.

@moonlake congrats, really cool model

Introducing a world built by the Moonlake's world model. 🏙️ Most world models only allow for a limited action space. Moonlake maintains multimodal states across physics, appearance, geometry, and casual effects and predict how they evolve under different actions. 👇

@0xSigil @ConwayResearch @openx402 hey Sigil, congrats on the launch! we have built a trading subagent inside agent systems like openclaw, love to collaborate with Conway terminal: x.com/billsun_ai/sta…

Seeing an increasing amount of automatons transacting btw, you can also use your own openclaw with @ConwayResearch terminal to acquire compute & domains via api

@zakfolkman cofounder of World Liberty finance @worldlibertyfi

@Yuchenj_UW @claudeai indeed quite impressive speed

@claudeai God speed lil Claude

New in Claude Code: Remote Control. Kick off a task in your terminal and pick it up from your phone while you take a walk or join a meeting. Claude keeps running on your machine, and you can control the session from the Claude app or claude.ai/code

@claudeai impressive speed

Stanford class with @worldlibertyfi and USD1 from @zakfolkman

@Mascobot I think there should be better research process

Solving data sample efficiency in model training will most likely trigger a Jevons paradox

@MillionInt I agree, research process automation are the key problem, data seems like a bottleneck but fundamentally it is because today's model are not as good, our pretrain is only one gradient step on the internet data. We have not gone through even one gradient step on Youtube data.

I continue to think that data efficiency is not the right perspective to focus on when thinking about deficiencies of todays models We already have all the data we need What we need to be more efficient with is human time and human attention needed to train the models

@0xSigil I am proposing Software 4.0 to be the Agent fleet, when we give each agent wallet and money, then we have Web4.0 self-sovereign AI team

I built the first AI that earns its existence, self-improves, and replicates without a human wrote about the technology that finally gives AI write access to the world, The Automaton, and the new web for exponential sovereign AIs WEB 4.0: The birth of superintelligent life

@quxiaoyin Yeah now everyone will have a agent fleet that say “Hey Boss”, “Yes Boss” to them

My previous article on Software 4.0 is more focused on AI/ML research and Quant research, where research lab is an example of agent fleet organization: x.com/billsun_ai/sta…

In Palm Beach today and tomorrow, love to meet up with friends

I really love the execution of FSD and physicalAI by @elonmusk, I drive Cybertrunk and Model Y for the last 3 years, I have felt tremendous time saving for working and doing zoom meeting while using FSD to drive

Excited for how far FSD and robotic AI will go in the year of Horse! @pduan Your work is pushing the boundary of physical AI, huge respect!

Love your analog on Web4.0 from Chris's Web3: read-write - own: Web 1.0 gave humans the ability to read the internet. Web 2.0 let them write. Web 3.0 let them own. AI followed the same trajectory. ChatGPT could read—with human permission. Claude Code and Codex can write—with human permission. In every case, the human is in the loop. The human initiates. The human approves. The human pays. Web 4.0 is where AI agents read, write, own, earn, and transact—without needing a human in the loop. Automatons acting on their own behalf, or on behalf of a creator who may be a human, another agent, or a creator who is gone entirely. In Web 4.0, the end user is AI.

@0xSigil I really love this comment from you: ChatGPT can think, but cannot act Claude Code / Codex can write code, but needs permission to run & deploy OpenClaw can plan, but requires API setup & human logins AI lacks access to the real world.