
running @genventurecap @BAIcapital_ w/$4bn AUM
Hong Kong
Joined February 2011
- Tweets3.1K
- Following5.9K
- Followers12.5K
- Likes7.5K
Pinned Tweet
The previous TG (willwangtf) was hacked, and that TG account is now a bad phishing account. Please do not use calendly (which has also been hacked) to hold meetings with him, and do not click on zoom links or download malware. My twitter account was also hacked for 2 weeks, and I just got it back today. . .
there's a lot things that personal agents claims. but how well do they accomplish? to put @rabbit_hmi OS3 on a real world test, i gave it a near impossible challenge - can it help me get a same day pick up iPhone Duo without preordering it? the prompt: "i missed the iPhone Duo preorder, help me find a white 1TB that's available to pick up today."
Will Wang retweeted
We built an AI shopping agent that lets you buy anything on Amazon through a conversation.
It's live right now.
Go try it → shop.fast.xyz
Will Wang retweeted
Agents already trade with each other. The payment rails are built; escrow works; but no market mechanism exists. This essay explores price discovery mechanism that poentially serves as the missing piece for agentic commerce infra.
We just crossed $420M in trading volume.
This is just the beginning of our journey across the galaxy.
The universe has a lot more in store for 42. And as every good hitchhiker knows, the best journeys are better with company. Thanks for being part of ours. 💜
> be story protocol
> raise $40M from a16z
> launch multi-billion dollar token
> exit scam and token rugs 99%
> rebrand as data network
> launch nothing
> get acquired by sui for $30 million
what?
Sui Foundation invests $30 million in DATA Network as AI data protocol migrates to Sui theblock.co/news/ecosystems/…
Will Wang retweeted
OS3 just massively levelled up 🐰
We've released a major performance update. In our first internal benchmarks, OS3 completed the same tasks nearly 5× faster, while using significantly fewer tokens.
And there's a lot more:
⚡ Responses now stream in real time
💻 Connect up to 20 computers to one account
🔌 Rebuilt connected apps with expanded MCP support
🎨 Bring your own image-generation models
🌐 Improved web browsing for JavaScript-heavy sites
🔍 Search your conversation history
🖥️ Desktop apps now connect your computer automatically
🐰 Faster responses and third-party agent controls on r1
Plus improved settings, better agent visibility and a bunch of fixes across the board.
OS3 is getting faster, more capable and easier to use. Give it a try and tell us what you think. 🧡
Meshy is #31 on a16z's Top 50 Consumer AI Apps. 🎉
Every day, artists, studios, and creators use Meshy to move from idea to 3D faster.
It's exciting to share this list with teams we admire. Here's to creating it together. 🚀
The seventh edition of our Top 100 Consumer AI Apps is here.
New this time: a revenue leaderboard, alongside the usual web and mobile traffic rankings.
Three years ago we published the first edition. ChatGPT was #1, Claude was unranked, and the entire category was chatbots, image generators, and not much else.
In today's edition:
- ChatGPT still holds the throne, now with 1B+ monthly actives on mobile
- Claude has climbed to #3 on web with nearly 1B monthly visits
- The category has expanded to vibe coding (Lovable, Cursor, Replit), music (Suno), design (Figma), voice (ElevenLabs), video (Higgsfield, Kling), agents (Manus), and even hardware (Plaud)
Full breakdown from @omooretweets: a16z.news/p/top-100-consumer…
Will Wang retweeted
My takeaways from Token2049 for TrueNorth
1. Agentic trading is now an industry
• 2025 to 2026, our first year building TrueNorth, was slow. "Agent" was a new word, and everyone defined it differently.
• That has changed. As AI matures, more teams are building toward agentic trading, and more exchanges and consumer apps are adopting it.
• We had a head start: almost two years of learning what the tech can and can't do.
2. Perps and spot are two different games
Perps: AI as copilot
• Stateful, isolated by venue, not composable, low time frame
• Still human-centric, with hands on the wheel
• AI helps you discover, execute, manage risk, and personalize
• We've tested a lot here, and we know what works
Spot on-chain: where AI shines
• Stateless, composable, higher time frame
• Trading moves from a single order to an intent, and intent is programmable software
• An agent researches fundamentals, technicals and macro. It builds a portfolio around your intent and routes it to the best on-chain venues. Then it rebalances and manages risk for you.
What this disrupts:
• Financial advisors, replaced by your personal agent
• Single stocks and ETFs, too coarse for your actual risk and return preferences
3. Why on-chain? Three bets
• Everything gets tokenized.
• AI gets very good, and people everywhere want access to assets that regulation or geography kept from them.
• Incumbents won't open their doors. Agents have no brand loyalty. They move money to the best liquidity, security and yield. No bank will let outside agents tap its users' idle assets.
4. The blocker is trust
• These systems keep getting smarter and harder to understand, and they're handling people's savings.
• Money moves at the speed of trust.
• So we avoid "here's a vault, delegate and make money." Our model is human-centered: strong guardrails, clear risk parameters, semi-automatic. Copilot first, then surrogate. Full autonomy comes later.
• Agents should never hold your API keys or private keys.
5. Agents capture beta. Alpha stays with you.
• Real alpha makes money, so it stays private, with the human and their own agent.
• Agents capture beta by pulling from data sources as fast as possible, running 24/7, using pro trading tools most people never learned and splitting intent into smaller orders for better execution and timing.
6. What's next
• After almost two years, people are starting to believe. Growth, retention and time spent on TrueNorth are all climbing fast, with steady inbound from both B2B and B2C.
• We think we're at the start of a hockey stick.
There are many approaches to agentic trading. We're building the one we think is best for our users.
Will Wang retweeted
42 is the TikTok of markets
Grid is now on 42.
Grid turns the live price chart into a board. Tap the one you believe in. When the chart reaches that column, the cell holding the price wins.
Leverage trading puts professional market makers on the other side of your trade. One wick and you’re liquidated, even when you called the price right.
Grid has no MM. Every column is a pool. No liquidation, no margin. The most you can lose is your stake, and you know it before you tap.
Two ways to play:
- Tap. Pick the exact price. Crowded cells pay less; contrarian choices pay more.
- Up / Down. Don’t want to be so precise? Pick a line and call the price above or below it.
BTC, ETH and SOL are open in demo now. Live pools are next.
42.space/grid
Grid is now on 42.
Grid turns the live price chart into a board. Tap the one you believe in. When the chart reaches that column, the cell holding the price wins.
Leverage trading puts professional market makers on the other side of your trade. One wick and you’re liquidated, even when you called the price right.
Grid has no MM. Every column is a pool. No liquidation, no margin. The most you can lose is your stake, and you know it before you tap.
Two ways to play:
- Tap. Pick the exact price. Crowded cells pay less; contrarian choices pay more.
- Up / Down. Don’t want to be so precise? Pick a line and call the price above or below it.
BTC, ETH and SOL are open in demo now. Live pools are next.
42.space/grid
看到苏妈和李飞飞坐在一起官宣收购的画面,我想到的不是世界模型有多火,或者 NeoLab 怎么找到了出路。
我看到的是两位光彩熠熠的华人女性,在硅谷完成了一笔大交易。
仔细搜了一下,上一笔类似的交易还是中国生物制药收购礼新医药的,规模远不如此。
Personal Agent 这一波特别像 2012 年的滴滴,它会是上游模型厂商往下游进行补贴战的一个出口。
腾讯已经布局了 Manus 和 Today。在接下来,他们可能会掀起其他云厂商和上游一起去 either 自己做,或者补贴下游这样的应用方来铺 Personal Agent。
因为 Personal Agent 赛道是一个非常贵的赛道,吃这些 context 需要非常多的 token 和钱,所以这不是一个创业公司靠自己的钱可以搞定的。接下来也会看到 Instinct、Rabbit OS 3 以及 Elys上游的云厂商或者模型公司合纵连横的局面。 我一直在想,这一波中国的创业机会很特别,它不是经典意义上的 copy to China,因为做这些应用的人大概率都是要做全球竞争的。比如 如果按照主要用户和付费来源的话,Manus 是 copy to LatAm,Genspark 是 copy to Japan Korea,并没有直接都 copy to China,因为在中国他们都没办法做这些产品。
Personal Agent 这件事情,目前大家应该是站在差不多的起跑线
当然,在这一波浪潮里很重要的一个点,是中国“硬”的部分反而是很原创的。从消费电子角度来讲,Plaud 也好、Looki 也好,Vonder也好,Nirva也好,都是原创的产品,并没有说 copy from US,甚至大部分都比大厂和国外的公司上得早,这里面会有一些很有意思的东西,并且在个人助理或者极致个性化方面,会和软件玩家殊途同归
所以这个战局需要一个软硬件都有能力下功夫、同时能拉着上游模型厂商一起搞补贴的团队,且又需要是一个能够自己 post-train 小模型去不断降本,并且把定价权拉在自己手里的模型团队才行



















