科技推文精选 - 2026年6月8日

2026年6月8日科技每日简报

Today's top tech conversations are led by @AskYoshik, whose post about 'One thing I've learned from te...' garnered the highest engagement. Key themes trending across the top stories include model, anthropic, claude, music, https. The community is actively discussing recent developments in AI, engineering practices, and startup strategies.

1. AskYoshik (Group Score: 249.2 | Individual: 31.9)

Cluster: 11 tweets | Engagement: 19 (Avg: 23) | Type: Tech

One thing I've learned from tech: The best technology doesn't always win.

The company that owns the user usually does.

That's why OpenAI is trying to turn ChatGPT into a superapp before the IPO. Every model eventually gets copied.

Same thing goes with X, that's what Elon Musk is trying to do.

Getting millions of people to open your app every day is the hard part.\n\nQT @Polymarket: JUST IN: OpenAI is reportedly planning its biggest ChatGPT overhaul yet, aiming to turn it into a “superapp” ahead of the company’s IPO.

See 10 related tweets

@business: OpenAI is readying a major platform overhaul ahead of its highly anticipated IPO later this year to ...
@HedgieMarkets: 🦔OpenAI plans to turn ChatGPT into a "superapp" with coding tools, AI agents, and paid integrations ...
@Cointelegraph: 🚨 JUST IN: OpenAI is reportedly turning ChatGPT into an AI superapp ahead of its planned IPO. https:...
@Polymarket: JUST IN: OpenAI is reportedly planning its biggest ChatGPT overhaul yet, aiming to turn it into a “s...
@rezoundous: omg which SaaS getting killed next\n\nQT @Cointelegraph: 🚨 JUST IN: OpenAI is reportedly turning Cha...

2. TheStalwart (Group Score: 224.5 | Individual: 42.3)

Cluster: 9 tweets | Engagement: 585 (Avg: 77) | Type: Tech

Can someone give me a clear answer to why everyone wants their own custom silicon. I kinda get wanting to be less dependent on Nvidia, but then aren’t you just moving your dependency one layer up to tsmc? Or are there genuinely performance gains to be had with model-chip aligned?\n\nQT @itsclivetime: Personal update: I’ve decided to leave OpenAI.

I’m proud to have been part of the custom chip program and grateful to everyone I got to build with and learn from along the way. The density of hardware talent on that team is extraordinary, and I don't think there's a better chip design team anywhere. It's been a wild journey from second hardware hire, 2.4 years ago, to now, and I'm excited to watch these chips become one of the most important engines of AGI.

At the same time, I haven’t been able to shake the pull to climb a new mountain from the bottom again!

I joined @AnthropicAI this week because I was deeply impressed with the team’s talent, values, and ambition, and I'm already energized by the pace and intensity of the past few days. It’s time to build.

See 8 related tweets

@ns123abc: “Sir… Clive Chan… our top chip designer on the Broadcom project… just quit and joined Anthropic BEFO...
@VaibhavSisinty: The same path. Again. Tesla → OpenAI → Anthropic. 🤯

First Andrej Karpathy made the move. Now Clive ...

@scaling01: Every time an Anthropic, OpenAI or Google employee switches labs it goes viral, because it is the mo...
@teortaxesTex: The charitable reading of custom silicon is obvious (you get AGI, you get your "nation of electric e...
@zzxwill: OpenAI 前硬件工程师 Clive（第2位芯片工程师，在职2.4年）刚离职去 Anthropic，他爆料：

OpenAI 在自研 AI 芯片，这款芯片将成为 AGI 的重要引擎之一，计划 202...

3. teortaxesTex (Group Score: 185.7 | Individual: 38.2)

Cluster: 6 tweets | Engagement: 109 (Avg: 48) | Type: Tech

Bullish for Meta Liang Wenfeng personally annotated data for V3 taste is something you've got to demonstrate by example, and it's not just glamorous architecture designs.\n\nQT @GergelyOrosz: Just learned:

Software engineers used to do manual data labeling at Scale AI while Alex Wang was CEO. After he left, new leadership joined, and were HORRIFIED to learn this. Stopped it ASAP

Now at Meta, software engineers are assigned manual data labeling... see the pattern?

See 5 related tweets

@jerryjliu0: not sure i see the issue here\n\nQT @GergelyOrosz: Just learned:

Software engineers used to do manu...

@dejavucoder: all of us are gonna be data labelers inshallah\n\nQT @GergelyOrosz: Just learned:

Software engineer...

@ShriramKMurthi: My hunch is that they’re labeling software runs and related data that need high expertise. Using cur...
@yoavgo: Strong agree with Alex here\n\nQT @GergelyOrosz: Just learned:

Software engineers used to do manual...

@chris_j_paxton: There are so many stories like this, about how important unferstanding data quality is for model per...

4. markgurman (Group Score: 171.1 | Individual: 28.6)

Cluster: 8 tweets | Engagement: 1317 (Avg: 766) | Type: Tech

Tim Cook has been intimately involved in Apple’s AI comeback and AI product direction since Apple Intelligence flopped and the new Siri was delayed. He’s never been this involved in product development and features as CEO. https://t.co/4hgg1PiTQp

See 7 related tweets

@cryptopunk7213: tomorrow Apple (finally) becomes a serious AI company.

Siri will become the #1 ai agent simply by ...

@ShishirShelke1: Oh it was serious https://t.co/U5TB6ioSSD\n\nQT @markgurman: Tim Cook has been intimately involved i...
@business: Tim Cook’s final act as Apple’s chief executive officer is getting the company to take AI seriously ...
@markgurman: Power On: Around early 2025, Apple’s top executives other than Tim Cook held a secret meeting to dis...
@business: Tim Cook’s final act as Apple's CEO is getting the company to take AI seriously, writes Mark Gurman....

5. Hesamation (Group Score: 153.4 | Individual: 30.2)

Cluster: 7 tweets | Engagement: 27 (Avg: 534) | Type: Tech

Notion confirms that Opus 4.7 and 4.8 have degrading performance. This is a repeating pattern we're seeing so far:

Models degrade after a certain time
Users report issues on Reddit and X
Anthropic can't find the issue for weeks

Meanwhile, Anthropic's "AI builds itself" and other reports proudly claim that Claude is helping write the code and train the models significantly.

We saw the exact pattern with Opus 4.6. Claude Code users complained it's not as smart as it used to be, for 6+ weeks, until Anthropic found three bugs causing the issue.

And you have to remember Anthropic has virtually unlimited Claude access internally, they can spend all they want on tokens, and they have billions of dollars to do so.

It seems like we're watching one of the side effects of AI coding unfold in front of us, which is random problems emerging left and right, and nobody having enough touch in the codebase to troubleshoot them.\n\nQT @NotionStatus: Anthropic's Opus 4.7 and 4.8 models are experiencing degraded performance, which is causing a higher rate of failures for users selecting these models in Notion AI.

To mitigate impact, all Anthropic models have been disabled in the model picker and requests have been rerouted to alternative providers. Most users should now be able to continue using Notion AI with minimal disruption, though Anthropic-specific features remain unavailable.

Please refer to https://t.co/2jxfCbG6UZ for the details.

See 6 related tweets

@andersonbcdefg: RT @Hesamation: Notion confirms that Opus 4.7 and 4.8 have degrading performance. This is a repeatin...
@zerohedge: RT @NotionStatus: Anthropic's Opus 4.7 and 4.8 models are experiencing degraded performance, which i...
@bradmenezes: Notion proves that it is critical for enterprises to not buy AI apps from AI Labs directly.

This c...

@ShanuMathew93: Woah is this one of the first high profile apps/partners straight up calling this out? Sure it’s hap...
@edzitron: Da future of software baby\n\nQT @NotionStatus: Anthropic's Opus 4.7 and 4.8 models are experiencing...

6. MarioNawfal (Group Score: 143.5 | Individual: 62.9)

Cluster: 3 tweets | Engagement: 7410 (Avg: 467) | Type: Tech

🚨🇺🇸🇮🇱 The Pentagon espionage bombshell just got worse...

New reporting confirms Israel has been eavesdropping on the American peace negotiators themselves.

The targets named: Steve Witkoff, Trump's top negotiator with Iran. Elbridge Colby, the Pentagon's policy chief.

And Michael DiMino, who runs Middle East policy.

Israel wasn't just collecting on America generally.

It was listening to the specific men writing the deal Netanyahu opposes.

What's wild is the documented incidents read like operations against an enemy state.

American defense personnel in Israel discovered spyware surreptitiously installed on their phones.

In 2021, Israeli military intelligence officers were caught planting listening devices at DIA headquarters.

Last year, Shin Bet officers were discovered trying to bug a Secret Service vehicle.

A SECRET SERVICE VEHICLE.

One senior official described the Israeli collection effort against top U.S. officials as "unhinged."

And the rating in context: Israel's counterintelligence threat level is now higher than any other American ally and EVEN HIGHER than some adversaries.

Notice who got targeted.

Colby is the most prominent advocate of foreign policy restraint in the government.

DiMino runs Mideast policy. Witkoff writes the deal.

Israel's espionage maps perfectly onto the people standing between Netanyahu and the war he wants to continue. They bugged the peace.

All of this while Israeli officers sit inside CENTCOM with unprecedented access, while Congress moves Section 224 to link the two militaries' systems and data permanently, and while Huckabee negotiates the framework.

The Pentagon says the ally is spying at a critical level.

Washington's answer is to hand them the keys to the network. Please make it make sense...

Source: NYT

See 2 related tweets

@MarioNawfal: 🚨🇺🇸🇮🇱 Tulsi Gabbard allegedly received the Israel critical threat report 18 days ago, and resigned ...
@MarioNawfal: 🚨🇺🇸🇮🇱 Hold these three facts in your head at the same time, because Washington apparently can't...

...

7. steipete (Group Score: 108.5 | Individual: 47.0)

Cluster: 5 tweets | Engagement: 4210 (Avg: 335) | Type: Tech

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore.

You should be designing loops that prompt your agents.

See 4 related tweets

@natemcgrady: I created an agent to create loops to create agents to create loops to cre-\n\nQT @steipete: Here’s ...
@ShanuMathew93: Non-technical idiot guy here. What does this mean for the non-coder audience? I truly want to feel w...
@chrisalbon: What does this loop like in practice? Like Claude -p in a for loop? /goal?\n\nQT @steipete: Here’s y...
@Shashikant86: Seriously need to figure out who is copying whom

https://t.co/CRYzV0deXb\n\nQT @steipete: Here’s y...

8. BrianRoemmele (Group Score: 95.8 | Individual: 37.6)

Cluster: 3 tweets | Engagement: 612 (Avg: 301) | Type: Tech

BOOOM! WE DID IT!

BRAINWAVE TO REAL-TIME MUSIC AI!

It has been a life long decades quest to read brain activity and to convert it to words, and/or music, colors and/or images.

Today I am very excited to announce with the assistance from Mr. @Grok director of The Zero-Human Lab, we have solved brainwave to music and this is the absolute worse it will be.

We found the code using an array of NeuroSky toy chips and our software pipeline connecting to open source ACE-Step 1.5 and a highly modified LoRA model we built for this. The lyric version is in testing now.

This would mean that the model will interpret words from the brainwaves and music!

Today we have the music side done and the quality and genera will expand. The is the worse it will sound.

Your Brainwave Music™️will be cut into 2-5 minute pieces based on a number of factors.

The specimen below is from a dream/hypnogogic state I was in last night and I have a recording of my thoughts after the state. The music was made in real-time and GUIDED the dream state with known technology like binaural beats (not easy to hear in this clip) and word back masking.

This specimen below shows the interplay of my brain state to the music made by my brain and adjusted to produce profound insights. I solved a very difficult issue in this session with a new AI model.

IT FREAKING WORKS!

THIS IS OUR FUTURE OF MORE POWERFUL BRAIN FUNCTION!

Our goal is to produce a portable device you wear and will be able to give real-time audio and PEMF (skull region), ultrasound (temple region) to maximize creativity and remote viewing.

It is very early days but I wanted you to know first!

YOUR support of my X account, just by reading this and sharing it, subscribing to my X, buying me a https://t.co/vvaNPw980t, and becoming a member at https://t.co/uwusvOVZqZ supports this research.

I will open source this at some point and build a device ANYONE can own.

Thank you!

I love you.\n\nQT @BrianRoemmele: I had a lot of folks laughing at The Human Synapse decoder I built from toys that had the NeuroSky chip. Today with the TRIBE v2 AI model our open source pipeline to move it rapidly in a garage to Thought to Text platforms! https://t.co/4GGg9RWTby

See 2 related tweets

@BrianRoemmele: THIS MUSIC WAS MADE WITH MY BRAINWAVES WHILE I WAS SLEEPING!

Someday you will do this to, it will b...

@BrianRoemmele: RT @BrianRoemmele: BOOOM! WE DID IT!

BRAINWAVE TO REAL-TIME MUSIC AI!

It has been a life long deca...

9. 0x_kaize (Group Score: 92.9 | Individual: 37.1)

Cluster: 3 tweets | Engagement: 169 (Avg: 330) | Type: Tech

NVIDIA is giving away FREE ACCESS to 80+ AI models

and almost nobody noticed

through their hosted APIs you get MiniMax M2.7, GLM 5.1, Kimi 2.5, DeepSeek 3.2, GPT-OSS-120B and more COMPLETELY FREE

plugs straight into Hermes agent, Cursor, OpenCode

here's the setup:

1/ grab your API key at https://t.co/KH7CDVEQbE 2/ base_url = "https://t.co/nW06KLlSAv" 3/ api_key = "$NVIDIA_API_KEY" 4/ pick a model (e.g. minimaxai/minimax-m2.7)

these days so many projects offer FREE ACCESS - all you have to do is take advantage of it\n\nQT @0x_kaize: a startup that raised $25M just put MiniMax M3 inside their platform

this might become the BEST AI FOR JUST $10/mo

what @CreaoAI does:

1/ you describe a task in plain English 2/ the Super Agent builds it, runs it, schedules it 3/ it runs 24/7 - AI builds it, AI runs it, you steer

M3 adds: a 1M context window, native multimodality, and up to 15.6x faster decoding at long context

my agent's been running on M3 through CREAO since launch. never crossed 50% of the quota once just pick M3 from the dropdown and run

seems like I just found the best and cheapest AI subscription

See 2 related tweets

@0x_kaize: a startup that raised $25M just put MiniMax M3 inside their platform

this might become the BEST AI ...

@0x_kaize: RT @0x_kaize: NVIDIA is giving away FREE ACCESS to 80+ AI models

and almost nobody noticed

through...

10. eng_khairallah1 (Group Score: 88.7 | Individual: 42.8)

Cluster: 3 tweets | Engagement: 496 (Avg: 77) | Type: Tech

RT @eng_khairallah1: Anthropic engineer:

"You're not supposed to prompt Claude. You're supposed to build a system that prompts itself."

this is one of the best workflows I've seen in a long time

in this video he breaks down exactly how most people are using Claude:

the 14% you lose to CLAUDE.md before typing a word
the plugins that 95% of users have never installed
the caching setup that keeps it at 95% hit rate and almost free
why starting every chat from zero is the slowest way to use Claude

if you've been using Claude for more than a month and never left the chat window, you've been using one project when you could be running a team of them

instead of another show tonight, watch this

make sure to bookmark it before it gets lost in your feed

full guide in the article below

See 2 related tweets

@eng_khairallah1: RT @eng_khairallah1: Boris Cherny, the creator of Claude Code at Anthropic, just shipped 125 setting...
@eng_khairallah1: RT @eng_khairallah1: Anthropic engineer:

"You're not supposed to prompt Claude. You're supposed to ...

11. StockMKTNewz (Group Score: 85.7 | Individual: 34.4)

Cluster: 3 tweets | Engagement: 128 (Avg: 462) | Type: Tech

WATCH OUT

Apple $AAPL is hosting their annual WWDC event with a keynote by Tim Cook tomorrow at 1PM ET

Apple is expected to announce its BRAND NEW AI powered Siri along with other AI features https://t.co/AXPgd02uRA\n\nQT @StockMKTNewz: THE BIGGEST IPO IN HISTORY IS ABOUT TO HAPPEN THIS WEEK

Here is the full catalyst watch I just sent to readers of my free daily recap

Get ready for the week ahead ⬇️⬇️⬇️⬇️

https://t.co/CO999faZLX

See 2 related tweets

@StockMKTNewz: APPLE $AAPL HAD A SECRET CRISIS MEETING IN EARLY 2025 ABOUT AI

The topic: Apple Intelligence was a ...

@StockSavvyShay: $AAPL held a secret crisis meeting in early 2025 after Apple Intelligence underwhelmed and the new S...

12. BradGroux (Group Score: 85.0 | Individual: 57.5)

Cluster: 2 tweets | Engagement: 650 (Avg: 31) | Type: Tech

RT @sriramk: 🇺🇸🚀 SOME NEWS: I'll be leaving my role at the White House at the end of this month. After a break I’ll be working on helping tackle some of the large challenges facing America on AI (more on that later).

It is hard to express how big a privilege it has been to serve the American people and how grateful I am to have had the opportunity to do so.

First and foremost, it has been an honor to serve under President @realDonaldTrump . Without his leadership, we would not be leading in the AI race.

Second, I owe a lot to the person I’ve worked mostly closely with over the last 18 months - @DavidSacks . His continuing advocacy for America winning on AI has been and continues to be crucial.

Some key public accomplishments from last year I’m proud of

Architecting and publishing the American AI Action Plan - charting the course for America to win on AI and helping execute on that for the last year.
The AI acceleration partnerships to help American AI stack win globally.
The National AI Policy Framework for Artificial Intelligence executive order (forming the basis for working with the Hill this year)
Advocating for the American AI stack with our allies globally (the AI summits in France and India, state visits to the UK, the Middle East and more)

So what’s next?

The past 18 months have given me a front row seat to this critical moment on AI facing America and our allies. Whether it is energy, data centers or a clear path for Americans to experience the benefits of AI, there are many tough issues we all need to navigate together. I plan on building institutions that help tackle some of those challenges for America and its allies.

I want to thank many others who have helped along the way in the administration : Kevin Hassett, @mkratsios47 , CoS @SusieWiles47 , VP @JDVance , @StevenCheung47 , Sec Bessent, Sec Lutnick, Sec Rubio and @jacobhelberg , @USWREMichael , Josh Gruenbaum, Watson Fagan, Ryan Baasch, Jeff Kessler, Alexei Bulazel, DepSec Landau, DepSec Dabar, Will Scharf, Taylor Budowich, @JamesBlairUSA , @elonmusk and many, many others. You know who you are and I know I’ll continue to see you a lot more.

Most of all, I want to thank @aarthir on supporting everything and being part of this unexpected but amazing journey from last January. None of this would be possible without her.

This journey has been the privilege of a lifetime and shown me how special this country is and how it needs all of us to contribute in anyway we can - and I plan on continuing to do just that.

🇺🇸

See 1 related tweets

@teortaxesTex: Unexpected Well, best of luck to Sriram\n\nQT @sriramk: 🇺🇸🚀 SOME NEWS: I'll be leaving my role at th...

13. Vtrivedy10 (Group Score: 80.6 | Individual: 54.8)

Cluster: 2 tweets | Engagement: 494 (Avg: 59) | Type: Tech

imo there’s a pretty solid default recipe that everyone should use to optimize a system of Agent = Model + Harness

you should “train” both

Build v1 agent using a sensible base harness and some task specific prompting + tools
Harness Engineering using eval tasks that roughly match prod

this is often enough - most companies can get acceptable perf doing this. then they collect traces, mine them for patterns, and make slight tweaks from there

SFT using data collected from traces) or synthetic data. Often is good candidate for “distillation tasks” to train a cheaper model while maintaining existing performance
RL if you have the bandwidth and ability and desire to create environments and designing rewards that represents the tasks you want your agent to be good at. Push past the SFT behavior of “copying” data from existing model to pushing past in some dimension
Light harness engineering again to squeeze any more juice (ex: slight prompting) using the trained model that’s better at your task distribution

this loop will largely be productized as a general purpose recipe for building and improving agents

we’re still in the earliest innings of the world’s companies getting comfortable with steps 1-2 of this loop. Harness engineering will probably be the dominant way ppl will optimize agents

but i expect a large number of companies to onboard through this entire loop on some trial project of interest in the next year\n\nQT @Vtrivedy10: https://t.co/Lb2gtM9SNs

See 1 related tweets

@hwchase17: RT @Vtrivedy10: imo there’s a pretty solid default recipe that everyone should use to optimize a sys...

14. chhddavid (Group Score: 78.6 | Individual: 30.5)

Cluster: 3 tweets | Engagement: 3 (Avg: 39) | Type: Tech

so you're telling me Claude Code Opus 4.8 can now...

build an app
launch it
do email marketing
self-build features

without any human in the loop??

it's so over... https://t.co/BbA9kEk0m3\n\nQT @shipper_now: Think it. Say it. Done.

The average person spends 7.5 years & $100,00s to build one app.

That ends today.

Meet Shipper: The first AI agent that turns your commands into a complete business people kill to pay for.

RT + Comment "Shipper" to randomly get free credits for 30 days.

(must be following so I can DM you)

See 2 related tweets

@Shipper_now: 🚨 JUST IN: Free credits to build apps w/ Shipper Max.

Everyone should become a top 1% builder. We'v...

@chddaniel: this is absolutely terrifying...\n\nQT @shipper_now: Think it. Say it. Done.

The average person spe...

15. FT (Group Score: 72.0 | Individual: 39.5)

Cluster: 2 tweets | Engagement: 305 (Avg: 125) | Type: Tech

Ahead of a planned listing this year, the $850bn company intends to transform the chatbot into a 'superapp' that combines coding tools and AI agents, adding products that executives believe will generate more revenue: https://t.co/2kAwdY5Yme https://t.co/Kl4LHk1UyA

See 1 related tweets

@ns123abc: they're rebranding the company based on codex lol\n\nQT @FT: Ahead of a planned listing this year, t...

16. sairahul1 (Group Score: 72.0 | Individual: 46.8)

Cluster: 2 tweets | Engagement: 463 (Avg: 84) | Type: Tech

This is the best site on the internet to learn harness engineering.

Free. Completely.

Most AI engineers have never heard the term.

https://t.co/bwDbTTYsjM

Bookmark this site.

Then read this setup ↓ https://t.co/ddEP0XowXM\n\nQT @sairahul1: https://t.co/uRsQBBhTcH

See 1 related tweets

@sairahul1: RT @sairahul1: This is the best site on the internet to learn harness engineering.

Free. Complete...

17. codewithimanshu (Group Score: 69.4 | Individual: 38.8)

Cluster: 2 tweets | Engagement: 796 (Avg: 246) | Type: Tech

RT @codewithimanshu: My girlfriend asked why I was smiling at 4 AM.

Showed her the terminal.

What are all those green numbers?

$1,129. Made while she slept.

Doing what?

Nothing. Claude scanned 14,000 wallets, found 47 that never lose, built a bot that copies them.

She watched for 10 seconds:

+ $3.87 captured +$ 6.42 captured +$12.71 captured

It just keeps going?

Every few seconds. New line. New money.

How much did you start with?

$300. Now$ 1,429. Eleven hours. Asleep.

What does it do?

Buys at $0.48. Sells at$ 0.52. Pockets $0.04. Who wins doesn't matter.

That's legal?

Citadel does this on NYSE daily. 400 engineers. I have one screen.

She looked at the P&L curve. Never dips. Just climbs.

Can you make me one?

Setting hers up now.

She still doesn't get how it works.

The bot doesn't care.

You only need Claude + laptop + 1 hour/day.

Giving This Free for 24 hours. To get it:

Comment the word "Claude"
Like and Retweet this post
Follow me @codewithimanshu (so i can DM you)

Save this post. Deploy the bot this weekend. Start with $300. Scale on evidence.

See 1 related tweets

@codewithimanshu: RT @codewithimanshu: I lost my job yesterday.

Rent was due. No backup plan.

Then I remembered I st...

18. myrhex (Group Score: 67.0 | Individual: 30.8)

Cluster: 3 tweets | Engagement: 92 (Avg: 24) | Type: Tech

Elon just confirmed that the 1.5T Grok model is now going through reinforcement learning.

This is the final major stage after the base training and fine-tuning. Once RL is complete, the model will be ready for release.

We are getting very close to the big 1.5T upgrade that everyone has been waiting for. This one is expected to be a significant jump in reasoning and coding performance.

SpaceXAI continues to move extremely fast through the entire training pipeline. Solid progress.\n\nQT @elonmusk: @techdevnotes It’s going through RL right now

See 2 related tweets

@mark_k: xAI News: @elonmusk says the 1.5T Grok Build model is already going through RL (reinforcement learni...
@Daniel_Farinax: How close are we getting to AGI once the 1.5T Grok model drops? 👿\n\nQT @elonmusk: @techdevnotes It’...

19. codewithimanshu (Group Score: 65.9 | Individual: 33.8)

Cluster: 2 tweets | Engagement: 452 (Avg: 246) | Type: Tech

RT @codewithimanshu: Claude scanned every Polymarket wallet and GitHub repo overnight.

Woke me up with $3,582 in profit.

The edge was not predicting the future.

It was finding the 11 wallets and 4 patterns that actually work while everyone else bleeds.

Most traders are chasing alpha in the wrong places.

Claude found it in:

→ Wallet behavior patterns across 12,000+ addresses → Open source trading scripts with proven P&L → Position sizing that scales with edge confidence → Arbitrage windows that close in under 90 seconds

It built an agent that combined all four.

Deployed at midnight. Ran while I slept.

94 trades executed. 87% win rate. $3,582 profit from$ 200 starting capital.

No predictions. No guessing. Just math exploiting information gaps.

The setup takes 1 hour to deploy.

The agent runs 24/7 after that.

You only need Claude + device + 1 hour to start.

Giving This Free for 24 hours. To get it:

Comment the word CLAUDE
Like and Retweet this post
Follow me @codewithimanshu (so i can DM you)

Save this post. Build the agent this weekend. Start with $200. Scale on evidence.

See 1 related tweets

@codewithimanshu: RT @codewithimanshu: This guy used Claude to build a Quant Bot and made +$589,139 on Polymarket.

25...

20. aakashgupta (Group Score: 65.4 | Individual: 33.4)

Cluster: 2 tweets | Engagement: 20 (Avg: 50) | Type: Tech

ChatGPT didn't hit 1B weekly active users because of better reasoning.

It hit 1B because of image generation.

That distinction matters more than most AI PMs realize. The reasoning upgrades (GPT-4, o1, o3) made the product dramatically smarter.

But they kept serving the same 10-20% of the population who are knowledge workers. People who sit at computers, read text, write prompts.

The rest of the world works differently. In India, knowledge workers are under 10% of the working population. Hundreds of millions of people whose work happens on WhatsApp, in markets, in small businesses.

They were never going to engage with a text-heavy chat interface that apologizes for its knowledge cutoff.

Image generation was the first obvious use case that didn't require being steeped in reading three paragraphs of text. You describe something, you see something. That's a UI model that works for everyone.

The hidden insight from this episode: the ImageGen 2 launch had the biggest ELO jump of any model OpenAI has shipped. Abhi, a PM at OpenAI, walked through exactly what drove ChatGPT's growth trajectory and the pattern is consistent. Each step-change in the WAU curve corresponds to a multimodal unlock, not a reasoning upgrade.

This is the product question most AI PMs aren't asking. Who are the next 800M users? They're not knowledge workers who need better PRD drafts. They're people whose primary interface with technology is talking, viewing, and messaging. Search and images reached them. Voice and computer use are next.

Build for the 80%.\n\nQT @aakashgupta: People keep saying GPT-5.5 + Codex mogs Claude Code.

So I got OpenAI PM @AbhiMuch to show me exactly how to use Codex like a pro:

03:48 - What Codex unlocks for PM work 05:34 - How he built his growth dashboard 11:32 - How to build in Codex with Playwright 14:52 - Moving away from PRDs to Codex prototypes 21:23 - 3 automations running before his day starts 28:37 - WhatsApp computer use demo setup 33:05 - Codex takes action inside WhatsApp 37:00 - Building a 1040 tax filing app in Codex 43:42 - What drove ChatGPT to 1B WAU 47:18 - ImageGen 2, the biggest ELO jump of any model 59:26 - How to break into OpenAI as a PM

See 1 related tweets

@aakashgupta: The PM who writes PRDs is losing ground to the PM who ships prototypes.

This shift is moving faster...