NoraW

[ClaudeDevs @ClaudeDevs]

How do we automate business analytics with Claude?

New blog post covering our best practices for skills, data foundations, and evaluations when building agents to perform data analysis:

https://t.co/mfEJMAQFBU

[Berryxia.AI @berryxia]

Stanford 这堂 2 小时 AI 系统构建课，直接把 Claude 所有教程和 Prompting Thread 秒了！

强烈推荐：
“比你刷过的所有 Claude 教程都实用 10 倍”

里面讲的不是 prompt 技巧，而是 Stanford 真正教工程师如何从零构建可靠 AI 系统的完整方法论。

周末就刷这一个，绝对是你这周最有生产力的事！

我已经将其翻译为中英 文双语视频直接戳这里👇

[Alex Finn @AlexFinn]

Hermes Agent is an INCREDIBLE AI agent

It's better than OpenClaw in some key ways, but does it replace it?

In this video I cover what Hermes agent is, how to use it, workflows that will explode your productivity, and if it's the OpenClaw killer:

[NoraW @norawangyn]

@ChenYu79203469 @animalkyat 拎起来太搞笑

[Trustmebro @trustmetobro]

TurboQuant in plain English:

Think of a smart AI chatbot like ChatGPT or Gemini. When you chat with it for a while or give it a long document to read, it has to remember everything you said earlier. It stores that memory in a special notebook inside the computer called the key-value cache.

That notebook gets huge really fast. It eats up tons of memory (RAM) and slows everything down. On a phone, laptop, or even a normal computer, this means:

• The AI can only handle short chats before it chokes.
• It needs expensive powerful hardware.
• Responses get slow.

Google Research just released TurboQuant, a new compression trick that:

• Shrinks that memory notebook by at least 6× (sometimes way more).
• Makes the AI up to 8× faster.
• Does it with zero loss in accuracy (the AI is just as smart as before).

It’s like taking a giant photo file, compressing it to 1/6th the size with a perfect zip tool, and the picture still looks identical when you open it. No blurry edges, no missing details.

What this actually means for regular people:

• AI chatbots can now handle much longer conversations without slowing down or running out of memory.
• It works better on phones, laptops, and cheaper computers—no need for giant data-center GPUs.
• Future AI (including Google’s own models) will feel snappier and cheaper to run.
• The little animation in the tweet shows colorful bars (representing AI memory) getting neatly packed into a tiny grid. That’s exactly what TurboQuant is doing behind the scenes.

Bottom line: Google figured out a smarter way to make AI’s memory tiny and lightning-fast without sacrificing quality. This is the kind of behind-the-scenes breakthrough that will make AI feel way more practical in everyday apps soon. No magic, just really clever math that finally works perfectly.

[tobi lutke @tobi]

Context engineering.

[tobi lutke @tobi]

I think watching Memento by Christopher Nolan is the perfect preparation for a career as a prompt engineer. Naturally explains the concepts of knowledge knockoff, limited context windows, and dangers of hallucination.

[NoraW @norawangyn]

@omooretweets @ChenYu79203469

[Deedy @deedydas]

🚨 This was the BEST Google I/O that I can remember.

Google launched over 12 different insane things.

Here is every single one of the launches and the best tweets about them:

1/12

[Circle @circle]

The time has come. 👀

On March 26, Cross-Chain Transfer Protocol (#CCTP) will officially launch on @Solana mainnet!

Ahead of launch, we’ve added a new pre-mint address to enable CCTP to mint $USDC on Solana.

Learn more about how pre-minting works:
https://t.co/srzqkVCJ4E

[Jeremy Allaire - jerallaire.arc @jerallaire]

Super cool demo. CCTP enables developers to build apps where users don’t need the native fee token on a destination chain, as a portion of the teleported USDC can be used to source native gas tokens via swaps and allow the user to start transacting immediately. Radically simpler user bootstrapping into apps across chains.

[NoraW @norawangyn]

To complete a successful deal,the following is also important:
1. The product itself is robust
2. Get to know what your customer really need
3. Provide a customized plan
4. Show benefits that linked to 2

[NoraW @norawangyn]

What differentiates a good salesperson/persuader/strategic maker/: sell the solution, not just sell your product.

[NoraW @norawangyn]

为什么序言不能把作者的名字先写在前面呢？我认为先阐明序作者和书作者之前的社会关系，才能更好的序言中的内容

[NoraW @norawangyn]

Start my new life! 😆😆 #bu2021