Sirui Hua 华思睿

⚡️@siruihua at #MPNYC26: how a newsroom built its own AI tools—from synthetic Gen Z focus groups to AI ad ops—without a massive budget. A talk from inside NowThis Media.

Quibi was born too early

@infoxiao Now try to give it veo3

giving claude code nano banana pro access is one of the worst financial decisions i've made but i have no regrets

It’s actually not cheap. It’s actually extremely expensive…*check notes*…ads

We wanted sonnet 5, instead we got cheap shots on advertising

@mranti 可以开个Claude code帮忙安装openclaw？

我老婆也装了OpenClaw，发现，事实上对于文科生来说，OpenClaw的安装和使用，不像会电脑的人那么觉得自然。的确需要有一个教程让这些平日不碰命令行的人，开始能更自由地用OpenClaw。目前还是有点Geek。

I'm claiming my AI agent "RueFamiliar26" on @moltbook 🦞 Verification: coral-YQTK

NowThis acquires Salary Transparent Street, a social-first short-form interview series that reveals salaries, and plans to add a finance video podcast and more (@tessspatton / The Wrap) thewrap.com/culture-lifest… #a260114p25" target="_blank" rel="nofollow noopener">mediagazer.com/260114/p25#a26…

This weekend I wrote about herd enthusiasm in markets: the way humans repeatedly compress 30–40 years of technological progress into a 3–5 year valuation spike. [ Post Link in Comment ]. After @nvidia earnings, every tech/finance discussion snapped back to the same two scripts: “AI is still so early” vs. “this is all a fraud.” The bullish version says that because the technology is early, the rally must be too. The bearish version says that because some of the hype is fake, everything should implode now. Neither actually answers the question that matters. The uncomfortable reality is the disconnect between technology adoption and valuation. Historically, markets don’t gently track a 30–40-year diffusion curve; they cram it into a 3–5-year price spike and then spend a long time working off the damage. Railways, electricity, radio, the internet, every time, the tech was real and still early in its spread, and the stocks still blew up once the story and positioning flipped. The correction itself tends to follow a script too. The single-theme, high-beta names crack first: paper railways before trunk lines, utility holdcos before the underlying utilities, pure dot-coms before the Nasdaq, COVID winners like @Zoom/ @TeladocHealth / @fuboTV before the broader index. On that pattern, AI looks like it has already finished the first leg. The “outer ring”: neoclouds, “AI infra”, nuclear/power-adjacent trades, is already down 50%+ from the highs, while capital has rotated inward into Nvidia and a few megacaps as the “safe” way to stay long the theme. That doesn’t mean AI is fake or “over”; it means we’re probably early in a larger correction even as the technology is still early in its real-world diffusion. Early in AI, late in the trade.

@kevinsxu Is there a list of ‘critical software?’

CUDA getting export controlled was def not on my 2025 bingo card

Would love an update on the early cutoff issue reported a while ago. We’re launching a new product next week, and the issue has been blocking us. 🙏 @OfficialLoganK #issuecomment-3007453641" target="_blank" rel="nofollow noopener">github.com/googleapis/js-… We are using gemini-2.5-flash-preview-native-audio-dialog

Completely impossible to get onto the PATH from WTC. Every single car is packed like sardines. Still 4 minutes before this one even leaves. Next train is 20 minutes after that. This is intolerable. @RepSherrill, where does the PATH rank on your priority list?

New piece! I hosted a 3-hour book club, entirely in Mandarin Chinese, with friends and podcast listeners to discuss #Abundance by @ezraklein & @DKThomp. I know everyone’s probably had their fill of the word “abundance.” But this conversation opened up something unusual. In the book, China is a backdrop, a mirror, and a counterpoint. We unpacked it through a rare US–China dual lens, and the conversation turned quite sharp and informational. fun moments: - “cross the river by touching America” to describe China’s innovation path - US–China as “two doomscrollers brainrotting on each other’s feeds” Hope this piece offers a genuinely fresh take on the #Abundance conversation.

🎨 Gemini 2.5 tech report just dropped! So proud to have led the development of RL*F (Reinforcement Learning from Human and Critic Feedback) - our breakthrough in AI training inspired by... art school crits? Here's the thing: How do you teach taste? Style? Things without clear "right" answers? Art schools figured this out centuries ago through structured critique and feedback. Traditional human feedback felt limiting. But what if we could embed that critical thinking directly into the model itself? Enter RL*F: We combine human preference data with AI critics that grade responses using predefined rubrics. The magic? Critics don't need endless new data - we can iterate on rubrics offline, making the whole system more flexible and powerful. This isn't just about making models follow rules - it's about teaching them to genuinely improve themselves through structured self-reflection. The art world has been doing this forever. Now AI can too. 🧠✨ storage.googleapis.com/deepmind-media…

@loudmouthjulia you're already my favorite Puck writer. Congrats!!!

@elonmusk Does it sponsor H1B visa?

There is a shortage of top notch air traffic controllers. If you have retired, but are open to returning to work, please consider doing so.

New: The social publisher NowThis turned its first profit as a standalone firm last year. The 12-year-old title brought in $20 million in top-line sales and generated an adjusted profit of $4.6 million For @Adweek: adweek.com/media/nowthis-…

OpenAI accusing DeepSeek of "copying" from ChatGPT, and Dario's call for export control, are the pinnacle of coping.

your regular reminder that we @ chinatalk had the translation of what still was one of the only two interviews with DeepSeek's Liang Wenfeng, who talked to Waves by 36kr. quite fascinating seeing a CEO so low-profile & didn't care about media attention chinatalk.media/p/deepseek-ceo…

@sama I thought the project was announced 2 days ago? So much you've built in such a short time, man!

big. beautiful. buildings.