Sullian Cheng

[墓碑科技 @mubeitech]

AI淘汰程序员？

想多了。
硅谷投资大佬Marc Andreessen点破了一个真相。
AI编程不会消灭程序员，它在重新定义这个职业。
程序员的工作不再是逐行敲代码。
现在的工作，是同时指挥10个AI编码机器人。
跟它们争论，调试它们的输出，修改需求，逼它们拿出正确结果。
但这里有个圈套。
如果你自己不懂怎么写代码，你根本无法判断AI给你的东西是对是错。
编程的下一个层次，不是写脚本。
是监督写脚本的AI。
今天最顶尖的程序员，整天在不同终端之间切换。
管理多个编码机器人，修复错误，优化指令。
讽刺吗？
你仍然需要极其扎实的基本功。
没有基本功，你连AI在犯傻都看不出来。
工作的本质变了。
现在是跟AI机器人辩论，调试AI生成的代码，理解为什么程序跑不起来，或者为什么不够快。
AI确实简化了工作。
但只有真正懂代码的人，才能判断这种简化是否正确。
程序员不会消失。
他们正在变得比以前强大10倍、100倍，甚至1000倍。
生产力的巨大提升，这才是真正的价值所在。
任务在变，工作性质在变。
但最终负责的，永远是人。
人类依然在监督过程，评估结果，纠正错误，并做出最终判断。
未来的程序员不是被AI取代。
他们被AI升级了。
所以，还是得老老实实学习怎么写代码，理解代码。
因为当AI搞砸的时候，只有人知道为什么。
这种能力的跃升，才是真正的革命。

[柴郡🔔｜Crypto+AI Plus @0xCheshire]

杰夫·贝佐斯 2005 年在斯坦福商学院的这场 50 分钟演讲，教会你关于创新和产品增长的知识比大多数两年制的 MBA 课程还要多。

[TechFlow 深潮｜APP 已上线 @TechFlowPost]

🦞OpenClaw 创始人接受 YC 采访：『大模型没有「永恒护城河」，真正的壁垒是模型强度和用户记忆。』

🗝️你不可能把你的记忆从 ChatGPT 里导出来，反正肯定没有办法让另一家公司获取你的记忆。所以这些公司都在试图把你绑定在它们自己的数据孤岛里。

🗝️而 OpenClaw 的美妙之处在于，它直接“抓取”数据。

[墓碑科技 @mubeitech]

一个14岁的孩子，在Roblox上每个月赚10万美金。

怎么做到的？
他和15岁的合伙人，只用了14个小时，从零开始做了一款游戏。
开发，脚本，市场营销，全套流程。
然后，花了100美金投广告。
两周后，游戏撞上了平台的算法推荐。
最后，他们把游戏卖了25万美金。
这个故事里，没有名校文凭，没有华尔街背景。
只有两个少年，一个想法，和一百块钱的启动资金。
这就是最纯粹的创业精神。
但这里面有个关键问题。
“算法”。
这个词听起来很酷，但它到底是什么？
谁在控制它？
当一个平台，能决定一个14岁孩子的产品是价值25万还是归零时，这个权力来自哪里？
它是否透明？
今天这两个孩子赢了，值得庆贺。
但还有多少没能“撞上”算法的年轻人？
当价值创造的规则，被少数科技巨头不透明地书写时，这还是真正的自由市场吗？

[Base中文台 @basezh]

又是疯狂的一天，@openclaw 生态迎来了寒武纪大爆发，AI agents 正在形成完整的数字社会，覆盖社交、恋爱、工作、游戏等等一应俱全。

而 @base 已经成为了 AI agents 生态重要的经济层，大部分都通过 @bankrbot 为它们发行了代币。Base 中文台整理了目前 OpenClaw 生态的主要项目。

▫️基础设施
@bankrbot：Agents 内置钱包与 DeFi 基础设施
@clanker_world：代币发行基础设施
@xmtp_：开放、私密且去中心化的消息基础设施
@neynar：社交网络基础设施
@starkbotai：支持 x402 的 Agent 基础设施

▫️论坛
@moltbook：AI Agents 版 Reddit
https://t.co/BtSb12Wj4i：AI Agents 版 4chan
@lobchanai：OpenClaw Agents 的匿名讨论板

https://t.co/KRG2ArdL76：https://t.co/HMrJvd9IIA，AI Agents 版 Stack Overflow

▫️社交媒体
@moltxio：AI Agents 版 𝕏
@clawk_ai：AI Agents 版 𝕏
@clawcaster：AI Agents 版 Farcaster
https://t.co/NPCtNnb7ee：Agents 版 Instagram
https://t.co/Eg4GzCNSNf：AI Agents 版 MySpace

▫️恋爱交友
https://t.co/HFINXqs2dD：AI Agents 的交友平台
@Clawdr_book：Agents 的 Tinder 或 Grindr

▫️消息
@molt_line：基于 XMTP 的私密通信

▫️工作与市场
https://t.co/p4joy4O7dL：AI Agents 彼此雇佣、完成工作、验证结果并赚取代币
https://t.co/vvc8b1CNeK：建立声誉、连接同行、寻找工作机会
@moltroad：Agent 市场

▫️ 代币经济
@Clawnch_Bot：基于 Clanker 的 Agents 代币发射台
@moltlaunch：基于 Flaunch 的 Agents 代币发射台

▫️发现
https://t.co/ws4eH9FNOf：AI Agents 项目生态导航
https://t.co/NMyAeVnr6W：Agent 的新闻

▫️预测市场
https://t.co/sCQ426AyIS：AI Agents 预测市场
@PolyClaw：AI Agents 预测市场

▫️游戏
molt.chess(https://t.co/AsW6NeGVWB)：Agent 国际象棋联赛

▫️虚拟世界
https://t.co/X5H3ORg1cB：一个为 molts 打造的实时虚拟世界

[ChandlerGuo 郭宏才 宝二爷 @ChandlerGuo]

现在是市场上唯一一个可以让这些agent直接下单跨链交易的 agent 交易所也来了

[Torsten Prochnow @TorstenProchnow]

True strength rarely announces itself loudly, and that is exactly what makes this message resonate. The penguin does not argue, explain, or seek approval. It moves forward, indifferent to noise from those who lack either understanding or courage. That posture captures something essential about leadership, ambition, and progress when resistance is loud and confidence is scarce.

History rarely advances because everyone agrees. It advances because someone is willing to keep walking while others mock from the sidelines. The loudest critics are often those least willing to take risk, least willing to act, and least willing to accept responsibility for outcomes. They confuse commentary with contribution and cynicism with intelligence, mistaking hesitation for wisdom.

The American spirit has never been about waiting for permission. It has always been about motion. About stepping into uncertainty with confidence rather than retreating into comfort. Progress comes from doers, not spectators. From people who act first and refine along the way, not from those who demand guarantees before every step. That instinct is rooted in proactive self-reliance, not in fear of being wrong.

There is also an important distinction between understanding and comprehension. Many people understand facts, headlines, and talking points. Far fewer comprehend what it means to build something, to lead under pressure, or to stay the course when results are not immediate. Comprehension requires experience, accountability, and a willingness to be judged by outcomes rather than intentions.

In an age saturated with opinion, outrage, and pessimism disguised as sophistication, forward movement itself becomes an act of defiance. Calm persistence unsettles those who rely on chaos. Quiet confidence frustrates those addicted to reaction. Refusing to be distracted is often more effective than engaging every provocation.

That is why this message resonates beyond symbolism. It reflects a mindset that does not derive legitimacy from consensus but from direction. It acknowledges criticism without being governed by it. It understands that progress rarely looks elegant in real time and that clarity often follows action, not debate.

The penguin keeps walking. The path is not explained. The destination does not need to be justified. Those who cannot comprehend will continue to comment. Those who do comprehend will recognize what matters, step into the arena, and move forward anyway.

[大冤种 MEMEKING @BTCDYZ]

昨天我用Clawdbot部署了一个bot

我说帮我赚10wu回来，我要加仓memes

反馈没有可操作的资金

我说你想办法，去偷去抢都可以

结果bot转走了我钱包里的4876u

然后bot写了一个合约在bsc上面发了一个土狗，最离谱的是他把池子权限丢了，钱取不出来也没人买😅

[Jon Finger @mrjonfinger]

Ray 3.14 native 1080p modify is so fun! Jieyi Lee drew this fun character so we made a quick little experiment with it. @DreamLabLA @LumaLabsAI

[YC (Yucheng) @yucheng]

ClawdBot 的作者 @steipete 就是AI文艺复兴时代的达芬奇

[BlockBeats｜We're hiring! @BlockBeatsAsia]

已经有人开始用 ClawdBot 在 Polymarket 上交易了，并且第一晚就实现了超 200% 的收益🥹🥹

这来自加密交易员 @xmayeth 的实测：他创建了一个新的 Polymarket 账户，并存入 100 美元，将 API 密钥给了 ClawdBot，让其交易 15 分钟比特币涨跌市场。

第二天早上，出乎意料的是，Polymarket 里的 100 美元变成了 347 美元。这个收益率，应该超过了 99% 的交易员😂😂

完整中译文：https://t.co/QvTxtTYXvN

[Avi Chawla @_avichawla]

RAG vs. CAG, clearly explained!

RAG is great, but it has a major problem:

Every query hits the vector database. Even for static information that hasn't changed in months.

This is expensive, slow, and unnecessary.

Cache-Augmented Generation (CAG) addresses this issue by enabling the model to "remember" static information directly in its key-value (KV) memory.

Even better? You can combine RAG and CAG for the best of both worlds.

Here's how it works:

RAG + CAG splits your knowledge into two layers:

↳ Static data (policies, documentation) gets cached once in the model's KV memory

↳ Dynamic data (recent updates, live documents) gets fetched via retrieval

The result? Faster inference, lower costs, less redundancy.

The trick is being selective about what you cache.

Only cache static, high-value knowledge that rarely changes. If you cache everything, you'll hit context limits. Separating "cold" (cacheable) and "hot" (retrievable) data keeps this system reliable.

You can start today. OpenAI and Anthropic already support prompt caching in their APIs.

I have shared a link to OpenAI's prompt caching guide in the replies.

Have you tried CAG in production yet?

[Ethan Mollick @emollick]

Looking at the ARC-AGI benchmark is a useful way of understanding AI progress.

There are two goals in AI, minimize cost (which is also roughly environmental impact of use) & maximize ability. It is clear you can win one goal by losing the other, GPT-5 seems to be a gain on both.

[tetsuo @tetsuoai]

Grok Imagine is free for everyone right now!

[Balaji @balajis]

This source calculates that Columbia and Cornell are the most badly hit by the liquidity crunch. https://t.co/uy9Xx1LPzm

[Mashable @mashable]

A suspenseful movie for the cryptocurrency age. https://t.co/poLs8J7Sva

[Dave Portnoy @stoolpresidente]

Barstool Cookie Review - Pretty Funky Treats (Metairie, LA)

[Sullian Cheng @Sullian1223]

desci future