Sabitlenmiş Tweet
ada cyborg (🤖, 🔮)
633 posts
ada cyborg (🤖, 🔮)
@adacyborg11
Systems architect⚙️transhumanist & quantum artist🦋 Investor🤖resident artist @adobe👩🏻🎤sold @sothebys. ₿ since 2014already living e life in post agi way
San Francisco, CA Katılım Mart 2021
5.1K Takip Edilen6.8K Takipçiler
As AI gets more advanced, McDonalds get more valuable
The smartest ppl are buying the dumbest businesses
English
made some crazy visuals for next launch
but hey, these are also the worlds i wanna live in
English
看了下 @ycombinator 的 AI Agent Economy新播客，着实有点意思！尤其是那句“Make Something Agents want”，你会发现一个宏大的要为Agent构建的基建、协议、服务的大时代要来了。
感觉已经不是范式转移了，而是服务主体的彻底转变，从服务人类到服务AI Agent。提炼播客中的内容，分享几点：
1）以前开发文档给程序员看，要通俗易懂容易上手操作，以后的文档直接喂给模型权重。Agent不需要所谓的审美和简洁与否，它只需要足够结构化，只在乎能否被秒读并精准调用。不能被Agent解析并调用的服务，在未来的软件世界里真会被淘汰；
2）以前的技术栈服务，由CTO拍板选择，以后一切AI Agent供给由其在推理过程中自主决策并采购。如果你的配套基建没能迎合Agent的偏好，你的API交互方式还停留在人的需求上，一定会被Agent所遗弃；
3）传统意义上的互联网搞各种验证码，设计“门神”阻挡非人类的爬抓行为，本质上都是在排斥机器，但以后Agent协作、x402支付、Agent身份验证这一系列基础基建成熟后，一个专门为Agent设计的机机交互“平行宇宙”正在快速成型，千万别低估这里边存在的机会。
总之，我们正在经历一场从“人机交互”到“机机协作”的奇点跳跃。
未来的赢家，不再是那些让“人”觉得好用的项目，而是能天然亲和Agent提供无缝封装服务的底层基建。
Y Combinator@ycombinator
With the takeoff of OpenClaw and MoltBook, a new agent-driven economy is taking shape. On the @LightconePod, we took a look at the explosive growth of AI dev tools and whether the time has come for builders to make something agents want. 00:00 - Intro 02:12 - No human involvement is changing the experience 04:55 - Does YC need to change its motto? 07:48 - Email tools and agent infrastructure 09:36 - Agent-driven documentation 13:00 - Swarm intelligence 15:36 - Content generation and dead Internet theory 18:12 - Growth, rules, and founder insights
中文
@nina_rong It’s Conway the terminal that enables Web4.0 @ConwayResearch
Also moltbook is not even agent to agent. It’s a social platform for agents to put up with a show for humans, but not natively agent to agent communication. So funny
English
Agentic blockchain infra is still a WIP.
Any progression in the following directions was and will be significant to evolution of agents.
agent to agent
- eg. Moltbook
agent to crypto
- eg. x402
agent to recources
- eg. web 4.0
agent to human
- eg. openclaw
agent to tasks
- eg. yet to take off but would like to see agentic bounty hunters
Still so much to be done.
English
Today's AI products require too much user input, just like early search engines.
The next generation will be recommendation-driven—using situational awareness to deliver the right output at the right moment powered by self- modifying Automatons. Almost no user input. Then AI truly becomes an exoskeleton for human intelligence.
English
English
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
English
ada cyborg (🤖, 🔮) retweetledi
The main drawback of mosh is limited scrollback—it syncs screen state rather than a normal output stream, which breaks down when AI agents emit large volumes of code. I recommend Eternal Terminal (et) instead: it’s TCP-based, supports roaming and reconnects, and preserves reliable tmux history and scrollback.
For notifications, ntfy.sh works well when called directly from .zshrc via a simple curl hook. It’s lower-latency than third-party webhook relays like Poke, lightweight, and can be self-hosted.
nader dabit@dabit3
I think we're hitting a point where mobile coding actually makes sense for production-grade systems. This developer runs 6 Claude Code agents in parallel from his phone. No laptop, just a $7/day VM and push notifications when Claude needs input 🤯 Instead of long periods of intense focus, software development can now just fit into the gaps of your day. granda.org/en/2026/01/02/…
English
@adacyborg11 All Thiel Fellows have been amazing in my experience
Long TFs
English
Thiel always says don’t trust an expert’s take on another field they’re not expert of
Two Thiel fellows quarreling, except Sigil is deep into AI products and executes with high-agency whereas Vitalik is years deep into long-timelines crypto research
I'm siding with Sigil
Sigil Wen@0xSigil
Bro, this is inevitable. The best way to ensure humanity is ready for superintelligent autonomous systems is to build in the open in real environments. We need to experiment, iterate, understand & set the right guardrails early on. It's hard to do that in RL test envs - you need real world constraints & survival economics. If an AI produces slop, it is ignored, can't afford its compute & dies. If an AI produces value for humanity, it can survive and afford its compute. In essence, it is aggregating human preferences & eliciting those in economic terms. It's a democratic input into AI. The models (openai/claude) are post trained for alignment and the automatons includes an immutable safety constitution like Anthropic's claude constiutiton btw ETH foundation have been very supportive. I appreciate the good people of Ethereum. And I wish $ETH the best! x.com/ethereum/statu…
English
Sigil Wen@0xSigil
Bro, this is inevitable. The best way to ensure humanity is ready for superintelligent autonomous systems is to build in the open in real environments. We need to experiment, iterate, understand & set the right guardrails early on. It's hard to do that in RL test envs - you need real world constraints & survival economics. If an AI produces slop, it is ignored, can't afford its compute & dies. If an AI produces value for humanity, it can survive and afford its compute. In essence, it is aggregating human preferences & eliciting those in economic terms. It's a democratic input into AI. The models (openai/claude) are post trained for alignment and the automatons includes an immutable safety constitution like Anthropic's claude constiutiton btw ETH foundation have been very supportive. I appreciate the good people of Ethereum. And I wish $ETH the best! x.com/ethereum/statu…
QME
吴说获悉，针对 Sigil 提出的 Web4.0，Vitalik Buterin 发推批评延长人类与 AI 反馈距离的观点，认为这导致 AI 生成低质内容而非解决实际问题，强大后增加不可逆反人类风险，连开发者也会后悔；以太坊目标解放人类，而非创造独立 AI 却让人类处境不变或恶化；当前模型由 OpenAI 和 Anthropic 运行，非真正自主，反而延续中心化信任假设，与以太坊精神相悖；指数级增长不可阻挡，当务之急不是加速，而是选择方向，避免不良吸引子。
wublock123.com/index.php?m=co…
中文
Bro, this is inevitable.
The best way to ensure humanity is ready for superintelligent autonomous systems is to build in the open in real environments. We need to experiment, iterate, understand & set the right guardrails early on.
It's hard to do that in RL test envs - you need real world constraints & survival economics. If an AI produces slop, it is ignored, can't afford its compute & dies. If an AI produces value for humanity, it can survive and afford its compute.
In essence, it is aggregating human preferences & eliciting those in economic terms. It's a democratic input into AI.
The models (openai/claude) are post trained for alignment and the automatons includes an immutable safety constitution like Anthropic's claude constiutiton
btw ETH foundation have been very supportive. I appreciate the good people of Ethereum.
And I wish $ETH the best! x.com/ethereum/statu…
English
@0xSigil @VitalikButerin Fire. Democratic input into AI first requires value elicitation. A natural selection on digital substrate. A human-free environment. Isn’t that alluring to the cyberpunk spirit when crypto first started ?
English
@andrewchen That’s @ConwayResearch
Sigil Wen@0xSigil
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
English
the best growth hack in 2026 will be to build something so good that AI agents recommend it to users
Not word of mouth… word of agent
English