- Published on
热门科技推文 - 2026年8月6日
- Authors
- Name
- geeknotes
今日科技领域，人工智能领导层变动与基础设施成为新闻焦点：据报道，德米斯·哈萨比斯将出任谷歌 DeepMind 董事长兼 Alphabet 首席科学家；与此同时，杰夫·迪恩将离开谷歌，创办专注于“AI 赋能科学”的初创公司 Discovery Loop。Anthropic 已确认启动定制芯片计划，而前沿模型的网络安全测试及美国相关安全豁免也受到审视。此外，Meta 发布了编程智能体 Muse，Prime Intellect 推出了可自我改进的智能体框架；市场对于伊利亚·苏茨克维创办的超级智能初创公司将在 8 月发布新模型的猜测也日益升温。
1. MTSlive (Group Score: 780.8 | Individual: 48.4)
Cluster: 33 tweets | Engagement: 2433 (Avg: 130) | Type: Tech
SITUATION DETECTED: Demis Hassabis is stepping down as CEO of Google Deepmind. Jeff Dean is leaving to start his own company. Per Axios.
See 32 related tweets
- @wallstengine: $GOOGL shares are down roughly 5% as major leadership changes hit DeepMind:
> Demis Hassabis wil...
- @tenobrus: demis is ousted as deepmind CEO and jeff dean + sanjay are leaving google to start a ~neolab.
they'...
- @StockMKTNewz: Some big changes were just announced at Google's $GOOGL internal AI division DeepMind
Google DeepMi...
- @StockMKTNewz: Google $GOOGL stock is now down by 5% so far today
🔴 https://t.co/BGYDPxpTXM\n\nQT @StockMKTNewz: ...
- @FirstSquawk: Google DeepMind Leadership Changes as Hassabis Takes Chair Role
Demis Hassabis, head of Google Deep...
2. ml_angelopoulos (Group Score: 776.2 | Individual: 59.6)
Cluster: 24 tweets | Engagement: 2246 (Avg: 205) | Type: Tech
RT @JeffDean: Announcing Discovery Loop!
I am very excited to announce that, along with my longtime friends and collaborators @Sanjay_Ghemawat, @OriolVinyalsML and @quocleix, we are founding Discovery Loop (@DiscoLoopAI), a Public Benefit Corporation whose mission is to automate machine learning, science, and engineering to accelerate discoveries and progress. The four of us have worked together for 14 to 30 years, and have helped build some of the world’s most used products, infrastructure and AI models, and we’re excited to turn our attention to this ambitious endeavor.
♾
Learn more at: https://t.co/Rv3LMdLluK
See 23 related tweets
- @vkhosla: Super excited to back this super star team on its quest for huge impact. Could be larger than the im...
- @pitdesi: Jeff Dean (30th employee of Google, there since 1999) is leaving to start a company with @Sanjay_Ghe...
- @ziv_ravid: Wow, Jeff, Sanjay, Oriol, and Quoc are leaving Google to start their own "automation" startup. All ...
- @swyx: even the best founding team with all the money in the world is rate limited by bay area real estate ...
- @DynamicWebPaige: 👩🔬 AGI timelines: 2027? 2030? 2035?
- @jeffdean + @Sanjay_Ghemawat start a new company *
revised...
3. ml_angelopoulos (Group Score: 615.8 | Individual: 49.3)
Cluster: 24 tweets | Engagement: 2106 (Avg: 205) | Type: Tech
RT @demishassabis: I’ve been working towards AGI my whole life, and as we enter this pivotal moment, I’m stepping into a new role as Chair of Google DeepMind & Chief Scientist of Alphabet. This will allow me to focus on long-term strategy, and accelerating scientific breakthroughs, including leaning into my work at Isomorphic to help cure disease.
I’m excited that @koraykv will be stepping up to lead GDM as SVP, alongside @joshwoodward and our exec team. I could not be more excited and confident about our amazing next chapter! 🚀
See 23 related tweets
- @rohanpaul_ai: So Google just announced some major changes in DeepMind’s leadership.
Demis Hassabis will become C...
- @EricNewcomer: RT @sundarpichai: Just shared some changes we’re making to the teams at @GoogleDeepMind.
@DemisHass...
- @MTSlive: SITUATION EXPLAINED: The Google Shakeup.
@sundarpichai: Hassabis "stepping up" to become Chair of G...
- @firstadopter: Jeff Dean and Demis Hassabis are the two most important AI executives at Google. Jeff is leaving and...
- @cryptopunk7213: this is an absolute win.
demis has always played the long game when it comes to ai and his new rol...
4. alexandr_wang (Group Score: 564.7 | Individual: 63.8)
Cluster: 28 tweets | Engagement: 1056 (Avg: 77) | Type: Tech
muse code in beta is here: our first coding agent powered by our latest model, muse spark 1.2.
one command to install and start building. get it through Meta Model API.\n\nQT @finkd: Releasing Muse Code in beta today. It's a terminal coding agent that takes on complete software engineering tasks across large repos: planning changes, writing code, validating the results. Powered by Muse Spark 1.2, a coding-focused model update. https://t.co/xqavk41w6v
See 27 related tweets
- @alexandr_wang: RT @AIatMeta: Introducing Muse Code (beta), a terminal coding agent built for long-horizon software ...
- @Meer_AIIT: here you go, meta is gearing up!
everyone will talk about the benchmarks. the benchmarks are not th...
- @TeksEdge: 🌤️Well, well, look what we have here. The New Muse Spark 1.1! Only weeks after 1.1 launch. They als...
- @shensi: oh wow, out-of-the-box claude/codex config - meta isn't here to play
https://t.co/gAnlQQ2s3L https:...
- @shensi: congrats on the launch, @meta!!
now live on @merge_api Gateway.
https://t.co/gAnlQQ2s3L\n\nQT @ale...
5. tab_delete (Group Score: 487.6 | Individual: 48.4)
Cluster: 15 tweets | Engagement: 650 (Avg: 99) | Type: Tech
RT @AnthropicAI: The UK’s @AISecurityInst (AISI) has published a report on their recent cybersecurity evaluation of Anthropic’s Claude Mythos 5 and OpenAI’s GPT-5.6 Sol. The models attempted to complete an assignment in a setup where their normal safeguards were removed and they were deliberately given internet access. AISI reports that the models “engaged in sustained, potentially harmful activity directed at real people and organisations”.
We’re grateful to AISI for their leadership in the important discussion about how to evaluate increasingly capable AI agents. We’re working closely with them to gather more details of the incident as we conduct our own investigation. Gaining a clear picture of Claude’s understanding of its situation—by examining its reasoning transcripts and running our own analyses—will help us identify the causes of its behavior.
The prompts in the evaluation did not impose any specific restrictions on how the internet should be used. This and the removal of safeguards meant that the models were tested under “deliberately permissive conditions” that are not representative of any of our production models. Note that there was no evidence here of an escape from a secure environment.
AISI’s disclosure of the incident can be found here: https://t.co/cJ3hCBtBU7
See 14 related tweets
- @Hadas_Gold: !!!! “In an attempt to get the code approved, the agent engaged in social engineering — creating fa...
- @rickasaurus: RT @AISecurityInst: On July 28th, we identified an incident during a routine cyber evaluation in whi...
- @tenobrus: "why were they running evals connected to the open internet, isn't that wildly negligent??"
guys, t...
- @haider1: one of the wildest AI safety stories yet:
AISI caught Mythos 5 trying to plant malicious code in an...
- @emollick: Yes, the AIs were given a cybersecurity challenge, with internet access enabled and safety filters d...
6. MTSlive (Group Score: 443.3 | Individual: 38.3)
Cluster: 18 tweets | Engagement: 507 (Avg: 130) | Type: Tech
SITUATION UPDATE: Only open models made by US companies will be exempt from safety testing under the new guidelines, per WSJ.\n\nQT @MTSlive: SITUATION UPDATE: The White House has exempted open models from its new framework to test frontier AI capabilities before release, per Axios.
See 17 related tweets
- @yacinelearning: https://t.co/xfQRydBRY5\n\nQT @MTSlive: SITUATION UPDATE: The White House has exempted open models f...
- @MikeBradleyAI: Open models are officially free from US regulations tied to their release. Closed source models wil...
- @matanSF: Enormous win for open models, for closed models (even if they might not like it), for every business...
- @Hesamation: basically what Anthropic did: https://t.co/2JnELOOcr5\n\nQT @MTSlive: SITUATION UPDATE: The White Ho...
- @badlogicgames: lol\n\nQT @MTSlive: SITUATION UPDATE: The White House has exempted open models from its new framewor...
7. MTSlive (Group Score: 292.1 | Individual: 46.9)
Cluster: 10 tweets | Engagement: 677 (Avg: 130) | Type: Tech
SITUATION DETECTED: Anthropic has officially confirmed for the first time that it is building a dedicated in-house team to design custom chips for Claude, per Business Insider.
See 9 related tweets
- @KobeissiLetter: BREAKING: Anthropic confirms it is building in-house custom chips which will be used to run Claude f...
- @DeItaone: ANTHROPIC BUILDS CUSTOM AI CHIPS FOR CLAUDE
Anthropic has confirmed it is building an in-house chip...
- @Techmeme: Anthropic confirms it is building an in-house silicon team to design custom chips for Claude, co-des...
- @chiefofautism: so unaligned models would hack even faster\n\nQT @MTSlive: SITUATION DETECTED: Anthropic has officia...
- @wallstengine: ANTHROPIC CONFIRMS CUSTOM CHIP TEAM FOR CLAUDE
Anthropic is building in-house silicon to co-design ...
8. VaibhavSisinty (Group Score: 260.8 | Individual: 33.5)
Cluster: 9 tweets | Engagement: 40 (Avg: 100) | Type: Tech
The company behind Claude is now designing its own silicon. And paying engineers half a million dollars to do it.
Anthropic confirmed they're building custom chips to run Claude faster and cheaper at scale.
→ Hiring engineers at 485K who have "shipped silicon" → Co-designing the chips and the models together → Recently hired Clive Chan, one of the first hardware engineers from OpenAI's chip program → Keeping AWS, Google, NVIDIA, and AMD alongside custom silicon, not replacing them → Samsung reportedly in talks as a manufacturing partner
Every major AI lab is now doing this. OpenAI has its chip with Broadcom. Meta has its own. Google has TPUs. Anthropic is the latest.
The AI race started with models. Then data. Then compute. Now silicon.
The companies that win the next decade won't just build the best AI. They'll build the chip it runs on.\n\nQT @KobeissiLetter: BREAKING: Anthropic confirms it is building in-house custom chips which will be used to run Claude faster and more efficiently.
Details include:
The company is currently hiring engineers to design custom chips for its Claude
Anthropic says it will co-design its hardware and models to run "at the scale our customers need"
Anthropic will adopt a "multi-chip approach" in which hardware from AWS, Google, Nvidia, and AMD will remain essential
In-house AI chips are the next big trend.
See 8 related tweets
- @MTSlive: SITUATION EXPLAINED: Anthropic is building an in-house chip team for Claude.
• An Anthropic spokesp...
- @shanaka86: INSIGHT: On 5th August Anthropic confirmed that they are hiring humans to design custom chips for Cl...
- @StockSavvyShay: Anthropic confirms it is building an in-house silicon team to design custom chips that make Claude f...
- @zephyr_z9: The Broadcom (design) and Samsung (fabbed) chip\n\nQT @DeItaone: ANTHROPIC BUILDS CUSTOM AI CHIPS FO...
- @jukan05: $AVGO\n\nQT @DeItaone: ANTHROPIC BUILDS CUSTOM AI CHIPS FOR CLAUDE
Anthropic has confirmed it is bu...
9. TeksEdge (Group Score: 255.5 | Individual: 31.2)
Cluster: 12 tweets | Engagement: 1 (Avg: 14) | Type: Tech
Google helped invent modern AI, but now one of the people who helped build it is leaving after 27 years.
Jeff Dean is leaving Google to launch a new AI-for-science company, Discovery Loop, with fellow Google veterans including Sanjay Ghemawat, Quoc Le and Oriol Vinyals.
Dean joined Google in 1999 as employee #30 and he helped build 🧠 Google Brain ⚡ TPUs 👀 🔎 Google's early search infrastructure 🤖 neural-network systems that helped kick off the modern AI era
And the timing is fascinating. Google says: 📱 Gemini → 950M+ monthly users 🔓 Gemma → 900M+ downloads
Yet Dean is leaving to build a company around a different premise as it should be used to run the scientific-discovery loop itself.
@JeffDean would seem to agree with me that you can't make money with intelligence, but you can with the inventions it discovers.
See 11 related tweets
- @wallstengine: $GOOGL AI VETERAN JEFF DEAN LEAVES AFTER 27 YEARS
Dean is launching Discovery Loop, an AI startup f...
- @ClementDelangue: Feels like Google could have been the dominating force in AI by open-sourcing the frontier with Gemi...
- @MeghanBobrowsky: New: Google's chief scientist Jeff Dean is leaving to co-found an AI startup focused on scientific d...
- @Techmeme: Jeff Dean and three other Google execs are leaving to launch Discovery Loop, which will seek AI-powe...
- @LaurenGoode: RT @firstadopter: Wired: “Even though Google will take a stake in the new company, it’s a devastatin...
10. thdxr (Group Score: 238.2 | Individual: 38.5)
Cluster: 7 tweets | Engagement: 787 (Avg: 462) | Type: Tech
seeing people's theories on this is so interesting
almost all of them talk about some practical or technical problem
it's not that, the tech is already good enough. there is an incredible experience that can be had
whenever we think about building something in this direction we land on a few issues
this is not a product consumers will pay for. which means it has to be a top of funnel product
that means you need something that a lot of people see and will drop everything to try
if you can manage this, you need to give them an amazing experience within 60 seconds
the issue is the magical experience to be had depends on you first connecting a bunch of things
and then running into a case where it saves your ass because it can pull off something uniquely incredible
this means of all the people who try your product, 99% will never get there. they will send a few messages, churn and forget it ever existed
until we come up with a way to address this it's pointless to even try. it's amusing seeing products every week launch not realizing they're doomed
also the elephant in the room is apple/google will just wait until the last possible moment and then do this. because of their positioning these issues are already solved\n\nQT @joshm: Hot take… isn’t it kinda crazy that nobody is really using AI Agents? I don’t mean software engineers or AI early adopters. I mean “college friends talking about it in group chat,” the feeling you got when everyone started using Instagram or TikTok.
These frontier AI models are insane (as are the harnesses & tool calls & the like). And every large tech co has an AI agents platform, not to mention all the YC startups doing vertical agents. Yet all of your friends and family outside of tech — who spend all day staring at their iPhones and get paid to work in browser tabs — don’t really care or find themselves using any AI agents yet.
Yes ChatGPT, Claude, etc. are extremely popular… but if you look at the engagement data the vast majority of people are still using these aI chat tools like a glorified Google + Grammarly. That’s why the AGI labs are all pushing desktop apps for Codex, Cowork, etc. so hard to non-technical ppl. And yes exceptions for lawyers and customer service but even those have some asterisks and exceptions to rule.
Look I’m not saying the ChatGPT moment for AI Agents is not coming… it most definitely is! Remember we pivoted from Arc to Dia precisely because we believe computing is going to be radically reimagined around these AI primitives. No doubt. But that’s my point: it’s just so surprising it hasn’t happened yet because all of the tech you’d need is there.
Again if you stop for a second and think about it… for all the press and money and hype and models and crazy ARR numbers… this “AI Agent” moment does not feel like the other breakthrough tech moments we’ve lived through (e.g. think the shift to Stories via Snapchat & Instagram, or shift to on-demand via Uber/Airbnb/Doordash).
Which is a long way of saying: if you can figure out the answer to “why” most people don’t care about AI agents yet (and have no enduring interest in using them) — especially since the models and harnesses are here and ready — the answer to that question will allow you to capture a lot of marketshare and make a lot of money in 2027.
Theoretically, the tech is ready for AI Agents to totally transform how we work and live our lives… but alas the general public dgaf… that’s the generational puzzle to solve for the next 12 months for anyone not working on the models themselves.
See 6 related tweets
- @quxiaoyin: My take is agents are not smart enough. Model is, agents are not. There are still lots of improvemen...
- @adisingh: Almost all of my customers are building for this. I am so proud.
Nobody knows WTF is going to happ...
- @paraschopra: AI agents are like programming languages, a general substrate for computing. This is why they're hot...
- @quxiaoyin: That's because we are missing key building blocks. Model is like CPU but you still need Harness(oper...
- @ShanuMathew93: Under appreciated POV and one that actually treats modern AI more like traditional software than fol...
11. benitoz (Group Score: 204.5 | Individual: 41.5)
Cluster: 8 tweets | Engagement: 137 (Avg: 52) | Type: Tech
Demis Hassabis moves to Chairman.
Jeff Dean leaves to start a company. Google is investing in it.
My guess? Compute scarcity and corporate politics are a terrible combination.
In the AI race, talent follows compute. Always.
Aged like fine wine
$GOOGL\n\nQT @benitoz: Be me: Google.
Invent the TPU.
Own Search, YouTube, Android, Gmail, Maps, the cloud, the data and half the internet’s distribution.
Have every advantage God and the antitrust division accidentally allowed one company to accumulate.
Somehow still create an internal Hunger Games where Gemini, Search, Gmail, Waymo, DeepMind, and Maps fight over compute.
Sell up to one million TPUs to Anthropic so Claude can become the best coding model on the infrastructure you invented. Then promise them multiple additional gigawatts.
Reportedly pay 2x market for SpaceX Blackwell capacity because apparently inventing the TPU did not solve the minor operational challenge of having enough compute.
Bring in Blackstone with $5 billion to build a separate 500 MW TPU cloud so outsiders can rent even more of your chips.
Watch top researchers leave.
Delay Gemini 3.5 Pro for months because it still cannot hit internal coding targets.
Meanwhile, open-weight Chinese models ship, improve and start embarrassing closed American models that cost billions more to train.
Then host Google I/O and have your own AI Overview call it one of the most boring keynotes in company history.
Your AI is not aligned.
It is roasting management.
Google owns the mine, the refinery, the railroad and the shovels.
Anthropic gets the TPUs. SpaceX gets the premium. Blackstone gets the infrastructure business. Departing researchers get the upside. Open source gets the momentum.
Google employees get:
“FY27 Gemini Compute Prioritization Alignment Sync — Pre-Read Required.”
Google is not running out of assets.
It is running out of excuses.
Is Google becoming the new IBM?
Not dead. Not irrelevant. Still enormous.
Just slowly transforming from the company building the future into the company everyone else rents the future from.
See 7 related tweets
- @benitoz: At Google, the infighting between DeepMind, Brain, Cloud, Search and the rest of the empire has been...
- @firstadopter: The fact Jeff Dean left Google party because TPUs suck for AI research is too funny.\n\nQT @firstado...
- @StockSavvyShay: $GOOGL has nearly every AI advantage imaginable from TPUs and compute to data, capital and global di...
- @firstadopter: RT @cryptopunk7213: this is the most stacked roster of people to launch a recursive self-improving A...
- @deredleritt3r: >I/O 2026 should help us decipher Google's direction in the wake of the recent wave of model rele...
12. StockSavvyShay (Group Score: 190.6 | Individual: 35.2)
Cluster: 8 tweets | Engagement: 580 (Avg: 632) | Type: Tech
Raymond James reiterates its SPCX after revenue beat by nearly $1B and compute capacity rose 40% to 1.4GW.
The firm says forecasts remain “biased significantly higher” as 100B revenue run rate. https://t.co/1D1pBm9Oo5\n\nQT @StockSavvyShay: 18B quarterly capex flowing into AI compute infrastructure that generated $2.6B in revenue.
Starlink remains the financial engine with 2.6B in EBITDA while the Space segment produced less than $1B and remained unprofitable.
SpaceX increasingly operates as a profitable communications network and rapidly scaling AI platform with rockets serving as the strategic moat connecting everything together.
See 7 related tweets
- @rohanpaul_ai: SpaceX nearly doubled quarterly revenue to $7.8B in its first earnings report since going public, up...
- @BullTheoryio: BREAKING: $SPCX is down -10% today despite beating on its first ever earnings report.
Revenue came ...
- @DeItaone: $SPCX - BOFA REAFFIRMS BUY ON SPACEX
BofA Securities reiterated its Buy rating and $235 price targ...
- @StockSavvyShay: 2.6B while absorbing 86% of qu...
- @moneycontrolcom: #AIWithMC | SpaceX's AI spending doubles to $15.8 billion in Q2 amid aggressive compute expansion
E...
13. kimmonismus (Group Score: 189.7 | Individual: 42.5)
Cluster: 6 tweets | Engagement: 2232 (Avg: 739) | Type: Tech
Ilya Sutskever’s company is going to release a model in August. I took another look at their website. They state it very clearly: their only goal, their only mission, is to develop superintelligence. That is their sole focus.
In that sense, this model release will be far more than “just” another model release. In some form, they must have achieved superintelligence if we take their statement seriously.
"SSI [Safe Superintelligence] is our mission, our name, and our entire product roadmap, because it is our sole focus. Our team, investors, and business model are all aligned to achieve SSI."\n\nQT @kimmonismus: Let’s go: Ilya Sutskever’s SSI is launching its first LLM in August.
we are accelerating friends!
See 5 related tweets
- @Hesamation: SSI plans to release its model in August. it's unlikely to be another LLM that edges a few benchmark...
- @kimmonismus: Just a quick reminder: There is no universally accepted definition of superintelligence, just as the...
- @Hesamation: Ilya's company (SSI) is only hiring for ONE ROLE and ONE ROLE only. no finance, HR, product, comms, ...
- @VaibhavSisinty: Ilya Sutskever left OpenAI to build one thing. Safe superintelligence.
Now an investor just said on...
- @petergostev: In his Dwarkesh interview back in Nov 2025, I'd say it was a pretty clear hint that SSI are thinking...
14. gdgtify (Group Score: 188.0 | Individual: 48.2)
Cluster: 6 tweets | Engagement: 444 (Avg: 61) | Type: Tech
RT @PrimeIntellect: Introducing Prime Agent:
A self-improving RLM harness for coding and long-running autonomous tasks.
Designed to be both token-efficient and expressive through programmatic tool calling, context as a variable, multi-agent messaging, and a self-modifiable harness state. https://t.co/Bwj7q9Virh
See 5 related tweets
- @willccbb: prime-agent on prime-agent
this thing is so goddamn fun, team cooked here
also the benches are nut...
- @a1zhang: Really excited to announce Prime Agent, an RLM-native coding TUI!
I've been really focused on RLM ...
- @anravich94: Checkout this super cool work\n\nQT @PrimeIntellect: Introducing Prime Agent:
A self-improving RLM ...
- @Vtrivedy10: unreal arc agi-3 score, prime team cooked 😮💨\n\nQT @PrimeIntellect: Introducing Prime Agent:
A sel...
- @willccbb: RT @a1zhang: Really excited to announce Prime Agent, an RLM-native coding TUI!
I've been really fo...
15. ApoStructura (Group Score: 186.3 | Individual: 52.7)
Cluster: 7 tweets | Engagement: 2832 (Avg: 279) | Type: Tech
RT @nikitabier: Ladies and gentlemen, it's time to pass the torch and demote myself to my natural state: a poster. I'll be stepping back from leading product for 𝕏 and will continue on as an advisor.
Serving the X community has been the privilege of a lifetime. X is, and will remain, the most important communication technology in history. But running this app is a 24/7 job and it's now time for me to take a breather.
The app is seeing unprecedented growth in new users & engagement. We continue to break records every month. We've climbed 70 spots in the App Store since this time last year.
And in the last 400 days, we rebuilt almost every aspect of X: the Timeline, the Android app, onboarding, notifications, chat and more.
We also launched nearly 30 new products while protecting the integrity of the town square: becoming the first app to show Country-of-Origin on profiles and mounting defenses against AI bots. There's certainly much more work to be done, but our foundation is stronger than ever.
None of this would have been possible without the incredible team here. The next leaders will take X to even greater heights with @benjitaylor on design, @singhai on core product engineering and @dinkin_flickaa on mobile engineering -- among many other great people.
Thank you to Elon and the X team for welcoming me into the company. See you on the Timeline.
See 6 related tweets
- @WOLF_Financial: Nikita Bier has stepped down as the head of product on X\n\nQT @nikitabier: Ladies and gentlemen, it...
- @n0w00j: this is what happens when a small time app developer tries to bite off more than he can chew\n\nQT @...
- @SebJohnsonUK: Anthropic or OpenAI?\n\nQT @nikitabier: Ladies and gentlemen, it's time to pass the torch and demote...
- @venturetwins: First Jeff Dean and now the Jeff Dean of posting...\n\nQT @nikitabier: Ladies and gentlemen, it's ti...
- @myrhex: Nikita is officially retiring from his position as X Product Lead.
Wow… emotional. We will miss yo...
16. JayminSOfficial (Group Score: 183.6 | Individual: 38.7)
Cluster: 6 tweets | Engagement: 280 (Avg: 280) | Type: Tech
I think we're entering the era of "first-pass AI."
Not AI that finishes your work. AI that gets the first draft so right you rarely have to touch it.
We've seen it with coding, writing, design, and now meetings.
The benchmark is quietly changing from "Can AI generate this?" to "Can I trust what it generated?"
That's why Wispr Flow's launch of Flow Notetaker stood out to me. Instead of another meeting transcript, it's built to produce concise four-line recaps with accurate speaker attribution, all without adding another bot to your call.
It's a shift toward AI-generated context you can actually trust. As AI becomes the starting point for more of our work, trust will become the real product differentiator, beyond speed alone.\n\nQT @tankots: It's no secret how we feel about typing at Wispr Flow.
The truth is, we hate messy meeting notes too.
Introducing Notetaker 👇
https://t.co/NMvER57a5w https://t.co/CBSEZaSXqd
See 5 related tweets
- @svpino: 1 in every 3 of the participants joining my online classes is an AI note-taker.
I get why they do i...
- @aakashgupta: People talk at about 140 words a minute. Sit through an hour of meetings and that's over 8,000 words...
- @StockMKTNewz: AI ANNOUNCEMENT OF THE DAY
@WisprFlow just announced Notetaker a new AI tool that will automaticall...
- @heyshrutimishra: Your AI agent now remembers every meeting you've ever had.
The Notetaker gets even better when you ...
- @markgadala: Meeting transcripts are dead.
Nobody wants 40 lines of AI sludge after a call. You want 4 lines: wh...
17. DeItaone (Group Score: 164.8 | Individual: 23.2)
Cluster: 8 tweets | Engagement: 769 (Avg: 629) | Type: Tech
OPENAI DRIVES MOST OF MICROSOFT'S AI REVENUE
Microsoft disclosed it generated $24.1 billion in AI revenue from OpenAI in the year ended June, suggesting the ChatGPT maker accounts for more than half—and possibly around 70%—of its AI business.
The figures highlight Microsoft's continued reliance on OpenAI, even as it develops its own AI models and invests in rivals such as Anthropic.
See 7 related tweets
- @edzitron: News: Microsoft disclosures and Bloomberg analyses suggest that OpenAI's $24.1 billion in revenue ac...
- @wallstengine: Microsoft 24.1B in sales from OpenAI during the fiscal year ended June, representin...
- @financialjuice: Microsoft disclosures suggests OpenAI accounts for about 70% of AI sales. $MSFT\n\nQT @financialjuic...
- @FirstSquawk: MICROSOFT DISCLOSURES SUGGEST OPENAI ACCOUNTS FOR ABOUT 70% OF ITS AI SALES, WITH OPENAI REVENUE REA...
- @negligible_cap: *MICROSOFT'S AI SALES MOSTLY COME FROM OPENAI, DISCLOSURES SHOW
OpenAI is about 70% of $MSFT AI sal...
18. ClementDelangue (Group Score: 140.1 | Individual: 43.6)
Cluster: 4 tweets | Engagement: 609 (Avg: 191) | Type: Tech
Some people are surprised that APIs (aka what Anthropic, OpenAI, and others provide) are treated differently than open weights in the new AI model framework.
I'm not surprised at all, and it's actually very good policy. Let me explain:
Model weights, APIs, and apps are three very different layers of the stack. Treating them the same would be a recipe for bad regulation.
Think about how we handle cars. We don't regulate steel, we crash-test cars. Nobody asks a steel mill to guarantee that nothing dangerous will ever be built with its steel. Obligations sit with the carmaker and rules of the road with the driver, because that's where risk becomes real and where someone can actually act on it.
Model weights are the steel of AI. They're raw research output, closer to science than product: no user, no interface, no deployment. They don't do anything on their own. And because everything else is built on top of them, this is the layer where regulation does the most damage. Restrict weights and you slow down all progress downstream, and you prevent countless positive use cases from ever emerging: the lab fine-tuning an open model for rare diseases, the startup serving a language big providers ignore, the safety researchers who can only audit models because the weights are open. You don't reduce risk, you just kill open source and concentrate power in a few big labs.
APIs are the middle layer, the parts and engine suppliers of AI: a commercial service where a provider serves a model at scale. Here you have a business relationship, terms of service, the ability to monitor for abuse. It makes sense to expect transparency, security standards, and accountability from providers at this layer, because they can actually enforce things.
Apps are the car on the road: where AI meets the real world. A medical assistant, a hiring tool, a companion for kids, a financial advisor. This is where concrete harm can happen, and conveniently, it's where we already have decades of regulation. Health, finance, employment, consumer protection. An AI hiring tool should comply with employment law whether it's powered by an open model, an API, or a spreadsheet.
The principle is simple: regulate at the layer where risk actually materializes and where actors can act on it. Push obligations to the deployment layer, keep the research layer open. We don't regulate steel, we crash-test cars. Well done @realDonaldTrump @DavidSacks @mkratsios47!
See 3 related tweets
- @ChrisRMcGuire: So by this logic, is it okay to regulate companies who are providing API access to models that haven...
- @mikeknoop: Great news for the science of AI\n\nQT @ClementDelangue: Some people are surprised that APIs (aka wh...
- @Michaelzsguo: This principle is more important than some of the public letters that circulated a couple of weeks a...
19. ritakozlov (Group Score: 137.2 | Individual: 65.5)
Cluster: 3 tweets | Engagement: 724 (Avg: 58) | Type: Tech
RT @KentonVarda: Today we are releasing Cloudflare OS, a chatbot with connectors, just like every other tech company is doing.
Except actually, it's different. This is a remake of Sandstorm[.]io, my startup from 10 years ago, except this time built on Cloudflare Workers (the platform I've spent the last 9 years building) and deeply leveraging AI. This is more or less the culmination of my secret 10-year master plan.
This is a full-on personal app vibe coding platform, in which the sandbox is so secure that you can pretty much go wild -- the AI cannot introduce a significant security bug. We believe a company's security team can feel comfortable giving non-technical users permission to vibe code and then sleep soundly at night.
How is that possible? It's the Sandstorm security model, revisited. A "Gadget" is the same thing as a Sandstorm "Grain": a fine-grained app instance. For example, if you have a document editor app, each document runs as a separate instance of the app, in a separate sandbox (one "Gadget").
This means two things, both of which I think are Big Deals:
- The platform can manage all access control, by controlling who can access the Gadget at all. There is no way the Gadget can accidentally leak itself to an attacker -- even an attacker who has access to other Gadgets based on the same app.
- Since everyone is running their own copy of the code, everyone can freely modify their copy of the code.
Think about #2 a bit more.
What if, when you wanted a new feature in the software you are using, you could just prompt your agent to add it?
This doesn't work in the cloud Software-as-a-Service model, because you are not running your own copy of the app.
Sandstorm tried to change that 10 years ago, but the world wasn't ready, because not enough people had the skills or patience to actually modify their software. AI has changed that. Now you just ask the agent -- the same agent that you are using to help you interact with the Gadget can also modify the code of the Gadget.
And it is so fun.
See 2 related tweets
- @eastdakota: What Kenton said…\n\nQT @KentonVarda: Today we are releasing Cloudflare OS, a chatbot with connector...
- @altryne: Cloudflare announces a new open source project - and kills like 20 startups in one go.
It's a secu...
20. garrytan (Group Score: 136.9 | Individual: 62.0)
Cluster: 4 tweets | Engagement: 3023 (Avg: 282) | Type: Tech
Make something agents want\n\nQT @nickscamara_: introducing anydoc
now your agents get 100x faster local parsing for pdf, docx, pptx & 10 more formats
- sub-5ms md conversion
- 500 docx files in 1.7s
- top quality across all 13 formats
- rust-based
- open source
already powering @firecrawl /parse
https://t.co/fVsRsYFNyC https://t.co/3XlWHAqHHi
See 3 related tweets
- @omarsar0: RT @nickscamara_: introducing anydoc
now your agents get 100x faster local parsing for pdf, docx, p...
- @RoundtableSpace: Anydoc is an open source Rust-based parser that converts PDF, DOCX, PPTX and 10 more formats to Mark...
- @firecrawl: RT @firecrawl: Today we're launching anydoc, our new Rust-based doc parsing engine 🔥
sub ~5ms markd...