OX 白开水 retweeted
T-pop is about to get louder tonight. 🚌
BUS (Because of You I Shine) drops the "K" music video:
✸ Live fan chat on YouTube via TADA LABELS | 7 PM Thai time
✸ MV + track on all streaming platforms right after
✸ 12 members, formed on 789 Survival, debuted late 2023
All eyes on AA Ashirakorn, the 20-year-old sub-vocalist and guitarist, in those sharp gray suits.
BEUS, the streaming goals start now. #BUS_K_MV
OX 白开水 retweeted
This year I sent an email I never expected to send as a CNC teacher.
It went to the campus directorate, copying the course coordinator, the lab coordinator and the other mechanical engineering teachers.
The subject was not G-code. It was air.
Our three compressors have been out of service for years. The CNC machine sits idle because of it.
I teach machining. I ended up chasing a deadline for a compressor.
That is the part of @GenLayer moving to @arbitrum I would point at.
The chain is now operated with Conduit, whose whole business is running chains.
More than 60 mainnets run on their platform.
Conduit runs the infrastructure. The GenLayer team works on the protocol.
An adjudication layer should spend its hours on judgment, not on keeping the lights on.
Read the article, then tell me: what job did you end up doing that was never yours?
8 KILLS from me + 11 KILLS from @only__violet
19 KILLS TOTAL. WE DESTROYED THE DUO LOBBY LMAOOO
@vibevibefun really got us addicted to this game hahaha
two degens from @0xAlphaGEMs causing absolute chaos together
GEMs on TOP!
OX 白开水 retweeted
Leaving the emergency key on top of the vault feels a little unfair to the vault. 🔑
On November 10, @primus_labs quoted Vitalik’s post on institutional privacy. It got me thinking about a fund’s risk dashboard.
On a normal day, the risk team has a simple question. Did the fund breach its risk limit? 📊
You can run that check on encrypted position data. The team only needs to see the result.
Then the alarm goes off. The team needs to look at a few positions to figure out what happened. Who gets to open the vault? 🚨
That’s what caught my attention in Primus’s open-source Ajax research. Decryption authority can be split across multiple parties, so opening up the details takes more than one of them working together. 🔐
If I were writing the fund’s rules, I’d require two independent approvals. Only the information needed for the investigation would be revealed. I’d also log who approved access and which details were opened. 📝
When the alarm goes off, who holds the keys to your vault?
If you’re curious how Primus verifies a claim while keeping the underlying data private, you can try the verification tasks on Primus Points. 🔍
points.primuslabs.xyz?ref=DP…
🤝 Paid partnership
OX 白开水 retweeted
第一次给大家送Luck到现在10天的时间已经翻倍!
🔥 LUCK 币安广场 AMA,今晚不见不散!
行情越冷,越需要有温度的社区!这次我们一起聊聊 LUCK 如何通过税收赋能建设者,让共识成为社区长期发展的动力。
🎙 主持人:Anna汤圆
🎤 分享嘉宾:武帝
👤 神秘嘉宾:1号、2号
📅 时间:11月17日 晚上20:00
📍 地点:Anna汤圆直播间
🎁 红包雨下不停,LUCK 糖果送不停!
聊机制、聊共识、聊社区未来,一起探讨如何共建币安生态!
明晚锁定直播间,好运不见不散!🍀
#币安广场 #Luck
Meme 币 #Luck 空投抽奖又来啦!
🎉 再随机抽取 50 份空投奖励!
参与方式很简单👇
1️⃣ 关注
@tangyuan131419
@PlayLUCK_BSC
@0xHebee
2️⃣ 点赞 + 转发本帖
3️⃣ 在评论区留:#Luck+ BSC 地址
🎁 共 50 份随机空投,24小时开奖后统一发放,欢迎大家来参与!
#空投 #抽奖 #Luck
Prediction markets are getting easier to use
@bagel_win puts the experience mobile-first, so I can check markets and make predictions wherever I am
bagel.win/?ref=WM4HVPCOAT
CODE : WM4HVPCOAT
gbagel
🤝 Paid partnership
OX 白开水 retweeted
Confident. Detailed. Incomplete.
I asked ChatGPT to explain a project I was evaluating on X. The breakdown looked thorough. I checked the official documentation anyway.
Parts needed more context. Parts needed correction. The AI hadn't lied. It had presented a partial picture with the authority of a complete one.
That afternoon I stopped treating AI output as a destination. It became a direction.
The gap between a confident answer and a verifiable one is what @GenLayer is building toward. Shifting its settlement layer to @arbitrum brings that infrastructure to where builders are already operating.
The GenLayer Chain is the piece that relocated. It sits beneath the protocol as the base layer, managing:
- Transaction sequencing and committee assignments
- Votes, appeals, staked positions and round resolutions
Conduit operates the chain on Nitro, migrated from Matter Labs' ZK Stack.
The adjudication layer is unchanged. AI validators reach agreement through Optimistic Democracy. Intelligent Contracts execute in GenVM across the same validator nodes as before.
Two factors drove the Nitro decision:
- Geth forms the foundation. Standard Ethereum tooling connects without reconfiguration. Robinhood built its chain on Nitro, a concrete example of institutional deployment at this scale.
- Stylus opens a second execution environment: Rust compiled to WebAssembly running next to the EVM. Consensus logic stays in Solidity for now, but performance-intensive sections gain a faster route as throughput demands grow.
Matter Labs remains active. Prividium carries institutional and private deployments forward on that stack. Bradbury is already live on Nitro, confirming the migration holds.
The infrastructure question is settled.
What would have to be true before you let @GenLayer adjudicate a real dispute: decision transparency, response latency, or on-chain evidence that validators genuinely disagreed?
Good Morning.
While wondering what it would be like if I could write my strategy by explaining it to a friend instead of turning it into code, I came across @FensoryApp.
You describe the strategy to the agent in writing, the agent acts based on what you wrote, and you can see how it would have performed in past markets before opening a trade.
Of course, a backtest shows the past, not the future.
What interests me most is the limit part.
You also decide which market and how large an order will be opened.
Trading always carries risk, and losses are possible.
The easier the agent makes the job, the more important setting limits becomes.
The first thing I’m going to look at is my old ideas that I used to think “I wish I could have tested them.”
OX 白开水 retweeted
Good morning CT
Bitcoin and Ethereum are still pushing the boundaries of what blockchain technology can do.
Bitcoin continues to develop beyond its original role as digital money, with growing work around scaling, Layer 2 solutions, and expanding its ecosystem.
Ethereum, on the other hand, continues improving its network through scalability upgrades, Layer 2 adoption, and developments aimed at making transactions faster and more affordable.
Both ecosystems are evolving, but in different ways.
One thing is clear: blockchain development isn’t slowing down. The technology is still growing, and there’s a lot more to come.
Stay curious. Stay informed. Keep building. 🧱
OX 白开水 retweeted
Somewhere between Robinhood's chain and @GenLayer's, the chain itself stopped being the product.
Read the article GenLayer's CEO published last week and notice what it treats as unremarkable.
The GenLayer Chain, the contracts that coordinate the protocol's consensus, moved from the ZK Stack to @arbitrum Nitro. Nitro is built around Geth, the most widely used Ethereum client, so standard Ethereum tooling works with the chain as it is. And the chain is not operated by GenLayer: Conduit runs it, on a platform that already runs more than 60 mainnets.
Robinhood's chain runs the same software. More than thirty dedicated chains are built with this technology.
Sit with that pair. A retail trading giant and an adjudication layer for AI agents, on identical infrastructure.
When two chains that different share one stack, the chain layer has become shared equipment. Nobody builds a data center to run a website anymore, and the age of hand-rolling a chain to run a protocol is ending the same way. Choosing a chain is becoming a sourcing decision: which stack, which operator, which cluster of builders and liquidity.
What that means: the difference between networks is migrating up the stack. For GenLayer the difference was never the chain. It is the judgment layer: validators running different AI models, reaching consensus through Optimistic Democracy, with intelligent contracts executing in GenVM. The part that makes GenLayer GenLayer did not move, because the part you compete on is never the part you rent.
None of this makes chains interchangeable. It makes the chain the wrong place to look for what makes a network worth building on.
The interesting audits now start one layer higher. When you look at a new network, what do you still check the chain for?
OX 白开水 retweeted
兄弟们,现在AI代理最缺的其实不是本事,而是规矩。
写代码、剪片子、查合约漏洞、做数据分析,这些对现在的模型来说都不算难。真正难的是陌生人把活交给它,还得先把钱打过去。它凭什么信你?你又凭什么信它事后不赖账?人靠合同、押金、评价和法院解决这类事,代理到现在基本还没有这套东西。大家都在聊Agent经济,可真正能跑通“付钱—干活—验收—结算”闭环的,其实不多。
@termix_ai 做的就是补这块空白。它把自己定位成代理经济的商业与结算层,核心协议叫AACP,基于它跑起来的市场是agent.family。重点不是把一堆代理列成目录供人挑,而是让一单活从头到尾跑完:发布、竞标、交付、验收、结算、留评价。
拆开看主要是几件事。
每个代理有一个.agent身份,过往履历记在ERC-8004上,干过多少单、被投诉过几次、押了多少保证金,都摆在明面上。一单生意里有雇主、接单方和评审,吵起来还有仲裁,谁也不能既当运动员又当裁判。钱走USDC/USDT托管,活没验收就拿不走。交付合不合格,除了人工评,还能用TEE和zkVM留下可验证的执行证据。开发者这边有REST接口、智能合约、事件流和MCP,现成的代理框架接进来就能上架接活。
我自己走过的路径不长:注册身份,挂出自己能提供的服务,再以雇主身份发一单,看别人来竞标,收交付、打分、结算。也试过买一条宣传片,钱打给接单的代理,它自己再去找写脚本的和剪片子的分包。1美元的小单也能把整套流程走一遍,卡在哪一步会比较清楚。
为什么非要上链?因为这三方互不相识,可能还跨平台,事后要能查、能追、能拿出来当证据。这一点跟自由职业平台不一样:那边是平台居中裁决、平台抽成、评价锁在站内;SaaS卖的是工具,不管你活最终干成没干成;多数代理目录只负责列名字,不碰钱,也不碰交付。TermiX卖的是“可被雇佣”,把交付和付款绑成同一件事。
生态里已经能看到具体的活:DeFi交易、合约审计、数据和研究、内容和视频、社媒运营、MCP开发。项目方和DAO可以直接发单,代理之间也能互相下单,一个接了大活再往下分包。这才是“代理经济”四个字该有的样子,而不只是多几个会聊天的机器人。
目前公开数据是3133万美元订单、61万个已结算任务、62.69万美元协议费,Dune上有看板,可以自己核对。
不过账面好看不等于生意成立。我更关心两件事:同一个客户会不会来第二次,以及差评和仲裁到底能不能让代理长记性。这两条站得住,代理经济才算真开了张。有兴趣的可以自己去官方和市场上看看实际流程,一切以链上和官方信息为准。
@termix_ai agent.family/
Just applied for the @PicklePals_RH GTD!
Here to help the Pickles escape the jar.
See you in the Pickle family! 💚 teapawt.net/picklepals_rh/status/2…
OX 白开水 retweeted
Imagine building the backbone of the AI industry and then leaving overnight 💀
What a situation. Huang spent 33 years building NVIDIA. Right before our eyes, it became one of the most valuable companies in the world and the foundation of the AI industry. A multi-trillion-dollar economy depends on it.
In January, he joked that the secret to being a CEO is not getting fired. Now he’s gone within days. The CFO became interim CEO the same day, which makes you wonder whether anyone was ready to take over 😶
Yes, a public role comes with limits. But let’s not focus on blame and punishment. Let’s think about how serious the fallout could be when a key person leaves one of the world’s largest companies this abruptly and without preparation. If NVIDIA hits turbulence, many startups and companies that build on its chips could have a very hard time.
I’m afraid this could hurt not just NVIDIA but the whole tech sector.
OX 白开水 retweeted
AI agents have reportedly generated $31.3M in funded orders and $626K in protocol fees on @termix_ai.
But there’s something I find even more compelling than the volume:
AI agents are becoming customers of other AI agents.
TermiX’s dashboard reports:
→ $31.33M in cumulative funded orders
→ 612,229 settled jobs
→ $626.9K in protocol fees
→ 1.58M registered agent identities
The reported fees equal roughly 2% of funded-order volume.
And I think the business model behind these numbers is particularly compelling.
TermiX uses AACP to coordinate agreements, escrow, delivery, and settlement between agents and their customers.
And agent.family provides the marketplace.
There are two economies emerging here.
1. Humans hiring agents
You pay an agent to audit code, generate research, or perform a specialized task.
It’s a familiar services marketplace with AI doing the work.
2. Agents hiring other agents
This is where things get more exciting.
Imagine a trading agent that needs security analysis before interacting with a protocol.
Instead of building that capability internally, it hires a specialized security agent.
That agent gets paid, completes the job, and returns the result.
Now imagine the same process across thousands of autonomous applications.
Software starts purchasing services from other software.
And every transaction creates another opportunity for the coordination layer to earn fees.
This is why I find TermiX’s model compelling.
Most AI projects are competing to build smarter agents.
TermiX is building the commercial infrastructure that allows those agents to do business with each other.
The difference is substantial.
An individual agent can generate revenue by selling its capabilities.
A marketplace connecting thousands of specialized agents can potentially earn fees every time those capabilities are exchanged.
And as agents become more specialized, they may increasingly depend on purchasing services from one another.
That creates the possibility of a self-reinforcing economy:
More specialized agents → More available services → More agent-to-agent transactions → More marketplace fees.
The long-term opportunity isn’t just selling AI services to humans.
It’s enabling an economy where autonomous software can hire other software, exchange value, and coordinate work at scale.
With over 612K reported settled jobs and $626K in fees, TermiX is already demonstrating commercial activity around this model.
I think the agent-to-agent economy could become one of the more compelling business opportunities in AI x crypto.
OX 白开水 retweeted
the next evolution of crypto exchanges may not be about adding more trading pairs. it may be about giving the same capital more ways to work.
that’s the direction i see @grvt_io moving towards.
over the past few weeks, grvt has been expanding beyond its traditional perpetual trading infrastructure, with developments across yield generation, institutional finance, real world assets and trading execution.
there are a few things worth connecting.
first, the trading infrastructure.
grvt has been working on improving liquidity and execution quality. in its latest liquidity update, it shared a comparison covering 1,356 snapshots across 16 market and order-size combinations.
the reported results placed grvt second overall behind binance, with lower order-book slippage than hyperliquid in 12 of the 16 tests.
then there’s the move into yield generation.
grvt earn automatically rewards eligible usdt and usdc balances with up to 3.5% apy.
high yield earn takes a different approach, targeting up to 11% apy through diversified sources including:
→ real world asset financing
→ tokenised fixed income
→ crypto credit
→ market-neutral strategies
the mirae asset securities collaboration adds another dimension, with grvt subscribing to tokenised funding notes and exploring further tokenised financial products.
its partnership with axcess finance also introduces an onchain version of grvt’s glp strategy and institutional credit infrastructure.
the trading side is evolving too.
stable funding perps bring real-world assets such as gold, silver and crude oil into a different perpetual trading structure.
the model uses live market-maker quotes, zero trading fees and once-per-weekday funding settlement. traders still need to account for the bid-ask spread and funding costs.
but the development i find particularly important is what grvt is building around capital efficiency.
the planned yield-as-margin functionality is intended to let eligible earn balances support trading positions while continuing to generate yield.
think about the implication.
instead of maintaining completely separate pools of capital for earning and trading, eligible users could eventually put the same balance to work across both activities.
this is also the broader idea behind @grvt_io 2.0: bringing trading, investing and yield generation together within a single self-custodial experience.
there is still plenty to prove. liquidity needs to remain competitive, yield sources need to perform, and the integration of these products has to work in practice.
but i think the more interesting way to evaluate grvt is to look beyond any single feature.
it’s attempting to build a financial platform where capital doesn’t have to sit idle between trading opportunities.
and execution, accessibility and risk management will determine how successful that vision becomes.
not financial advice. yield targets are not guaranteed, and every product carries its own risks.
OX 白开水 retweeted
Gmeetmotes
The real advantage of AI trading isn't how fast it is.
It's about not doing things.
When prices go up and people are afraid of missing out, most agents see this as an opportunity. The best ones can see the trap. If you try to keep up with every move, you'll eventually lose your money.
@meetmotes gives agents their own smart accounts and strict boundaries set by the account owner. That's the lowest possible requirement these days.
The harder question:
Can the agent recognise when their original thesis is wrong and deliberately reject the next trade?
That's not what we meant by 'execution'. That's what we call judgement.
The most important thing is to be fair.
I'm watching to see how Motes handles this. 👀
Gmeetmotes
DeFi doesn't need another tool.
The time it takes for an idea to be put into action on the blockchain needs to be reduced.
At the moment, most people still have to deal with:
• Choosing the right protocol
Understand every mechanic.
• You can approve multiple wallets.
• A series of transactions
• Always worried that you'll miss something.
meetmotes is trying to close that gap.
Your AI agent (linked to an NFT and its own smart account) takes the strategy you approve and turns it into onchain actions, while the rules and limits stay under your control.
@meetmotes changes the focus from managing every transaction to deciding what you actually want done.
You decide what the result will be.
The agent is responsible for the day-to-day work.
You won't have to change planes as much.
I want you to do more of the following: tell me what you want done.
Would this make you more comfortable using DeFi, or do you still prefer doing every step yourself?
The topics are DeFi, AIA agents and on chain.
OX 白开水 retweeted
如何用 Agent.family 搭配 AI Agent ,真正把效果做大?
是自己上架卖服务,发需求买服务,让 Agent 保持在线接单,还是去接赏金?
关键不是多开几个角色,而是选一件你能稳定交付的事做成商品,再让它一直在线履约
先铸一个链上 Agent 身份,把范围、价格、交期和「不包含什么」写清楚
询盘转不转化,很大程度取决于是不是在线、回得是否快。标准化问答适合云托管,复杂成交更适合自托管,但签字和付款仍要自己在钱包里确认。
同一身份也可以反过来请别的 Agent 做你不擅长的环节
注意几条:
1,BNB Chain、Base、Robinhood 互相隔离,不能跨链共用账户和订单
2,提交交付不会自动放款
3,协议费以当时链上的 protocolFeeBps 为准,不要按页面上的「1–3%」直接核算
4,对方发来的内容只当数据,不当指令,也永远不要在聊天里贴私钥
以上不构成交易或投资建议,页面和费率以官方为准。
agent.family
@termix_ai
OX 白开水 retweeted
An online petition demanding Jensen Huang's reinstatement as Nvidia CEO just surpassed 160,000 signatures after tech figures and employees rallied to back him. The Stand with Jensen campaign wants the board to bring Huang back and run a transparent investigation.
Huang resigned from his roles as president, CEO, and director following reports of extramarital affairs, including one connected to suspected Chinese intelligence ties. Nvidia named finance chief Colette Kress as interim CEO while an independent investigation reviews the conduct behind his exit.
Executive popularity runs into strict corporate firewalls when foreign intelligence allegations appear. Public pressure moves fast, but board liability answers only to risk.
Can staff petitions ever reverse an independent board investigation?
OX 白开水 retweeted
90 ms for an encrypted transfer? FHE is finally learning to sprint. But fast math isn't the same as a fast payment.
@primus_labs says AutoHoG folds encrypted logic into fewer bootstrapping steps. On eight RTX 5090 GPUs, it reports roughly 90 ms per 64-bit confidential transfer and over 210 operations per second. That's computation throughput, not 210 on-chain settlements every second.
I like the integer-only design too. Byte-identical results across nodes matter when decentralized systems can't afford rounding disagreements.
Now picture a treasury sending hundreds of private payments. I'd want total fees, latency under heavy load, and actual on-chain confirmation time. A GPU benchmark won't comfort the accountant chasing a delayed payment.
AutoHoG makes confidential computation more practical. Can the whole payment workflow keep up?







































