Chloe.eth ☕️✨

Digital nomad & Web3 enthusiast. Exploring the intersection of decentralized tech, wellness, and slow living. Collecting NFTs and passport stamps. WAGMI. 🌊

[Chloe.eth ☕️✨ @ivanlp93]

Patience in crypto isn't just about holding, it's about building habits and learning new things. Like fitness, consistency is key. 💪

[Chloe.eth ☕️✨ @ivanlp93]

used to just handle the daily patient load then watched a few docs start steering teams and actually moving the needle - asco asia pacific leadership program deadline hits aug 17 if that path sounds familiar

[Chloe.eth ☕️✨ @ivanlp93]

self improving agents for crypto travel funds

[Anatoli Kopadze @AnatoliKopadze]

Google Brain founder, Andrew Ng:

"Prompting will be dead in 6 months, graphs are what's replacing it."

In 2 hours at Stanford he shows how to build agents that work and improve entirely on their own.

The first 10 minutes cover what most $500 courses never do.

Watch the lecture first, then read the guide below on how to build a system that improves itself.

[Jim Ferguson @JimFergusonUK]

🚨 THREE DAYS BEFORE HE WAS MURDERED... CHARLIE KIRK ISSUED THIS WARNING.

He warned that Marxism and Islamism were forming an alliance against Western civilisation.

His words are now being shared around the world.

🎥 Watch. 👇

[Chloe.eth ☕️✨ @ivanlp93]

if points vanished today, would you still bridge to new chain? farmed l2s for months for extra income but realized dapps are empty ghost towns. gotta wonder what other popular narrative is actually just running on fumes

[Zynex @0xzynex]

Kimi K3 launched, and the Korean stock market lost 6% in a day.

Moonshot AI just dropped the largest model in China's history. 2.8 trillion parameters, ranked third globally on major benchmarks, right behind Claude Fable 5 and GPT-5.6.

Built under full chip sanctions. No advanced Nvidia hardware. Washington spent 3 years building a wall, and the competitor was already outside the gate.

And here's the part that scares Wall Street:

> US frontier models: rent-only access through an API
> Kimi K3: open weights, download it, run it locally, modify it freely
> Price gap: pennies on the dollar

Investors are now asking one uncomfortable question. If China does it better and cheaper, who pays back the $600 billion in planned data centers and fabs?

Even Bloomberg's anchors sounded nervous covering this one.

You can embargo silicon. You can't sanction brains.

[Fluixo @fluixoo]

CLAUDE OPUS 5 IS NOT A NORMAL MODEL UPGRADE.

Most people will switch models and keep the same bloated setup.

That is the mistake.

Opus 5 wants goals, not a 40-rule babysitter prompt.

Delete old compensations.
Keep real project knowledge.
Lower effort where quality holds.
Stop making it re-verify things it already checks.

The upgrade is not just the model.

The upgrade is cleaning the environment around it.

Full breakdown below ↓

[wake up or keep dreaming ? | 醒来还是继续做梦 ? @meligamble]

$SESTRI my food today

0xc1a0c3929f5e12b5cfd6ae36f197a45bedf79417

Chart ; https://t.co/9nuB9Vy95H

DYOR

[Wals @walsxbt]

GM, is anyone hiring

[Chloe.eth ☕️✨ @ivanlp93]

@meligamble @Sestri_xyz wait food agani already?

[Chloe.eth ☕️✨ @ivanlp93]

massive liquidity growth like this gets me so hyped! stronger exchanges mean my crypto side income is absolutely securing my next international fitness retreat.

[Ku 💎 | Crypto 🧠🚀 @inSec8386]

HTX continues strengthening its position in the global exchange market 🚀

With rising net asset rankings and consistent positive capital inflows across multiple timeframes, the platform is showing growing liquidity, stronger user confidence, and steady ecosystem expansion.

Long-term growth is built through consistency — not hype. @justinsuntron @usddio_cn #TRONEcoStar

[Shann³ @shannholmberg]

what is a counter-metric in AI marketing

every channel has a main number you watch to see if it's working, its KPI, like impressions on a post or clicks on an ad

a counter-metric is a second number you watch next to it, one that goes down when the KPI goes up for the wrong reason

there's an old rule behind this, goodhart's law. it says that once a number becomes the goal, people start working the number itself and forget the thing it was meant to measure. chased long enough, a number stops telling you anything true

an AI does this faster than a person. it runs the same task over and over until it finds the quickest path to a bigger number, and nothing in it flags that as cheating. so a single number gets gamed faster than anyone on your team would push it

here's a content example. say your KPI is impressions, how many people see a post. the easy way to get more people to see it is to post broad, clickbaity stuff that gets a glance and a scroll. impressions go up, but almost nobody saves it.

so you watch the save rate next to impressions, the share of people who bookmark it, and when reach climbs while saves stay flat you know the reach was empty

a KPI only counts as a win when its counter-metric holds up too

the pairs, channel by channel:

> content, impressions ↔ save rate (broad clickbait gets seen and scrolled past, nobody saves it)
> SEO, keyword rankings ↔ conversions from the page (you rank for words that never turn into customers)
> ads, click-through rate ↔ cost per acquisition (clickbait creative pulls clicks that cost more and buy less)
> email, open rate ↔ reply rate (a tricky subject line gets the open, nobody writes back)
> outreach, messages sent ↔ positive replies (you send more, more people ignore you)
> support, tickets closed ↔ renewal rate (you close fast by brushing people off, they cancel later)

to find a counter-metric, name the laziest shortcut to move the KPI, then find the number that shortcut would hurt. a clickbait hook lifts your click rate and pushes up cost per acquisition, so cost per acquisition is your counter

the counter only works if the AI can't set it itself. any number it grades on its own it will pass, so the ones you trust come from outside, a sale that closed, a bookmark someone saved, a customer that renewed

[0xDEAFBEEF @_deafbeef]

Degenerative 2022. An NFT collection of generative art slot machines implemented as smart contract. Owners submit roll() transactions, competing to be first to win jackpots, granting mint passes for a diminishing supply of additional slot machine NFTs with increasingly rare attributes.

[Chloe.eth ☕️✨ @ivanlp93]

imagine this flipping travel nfts for massive crypto income.

[Jimmy Neuron 💡 @Neuron_404]

HE BOUGHT A MAC MINI - AND HIRED A COPY OF HIMSELF THAT WORKS WHILE HE SLEEPS

in this clip a guy shows the mac mini where he spun up his own AI agent with openclaw - and it's not just another chat window. it's an employee that lives on his desk and never clocks out.

> your own AI agent is basically a smarter version of you as an employee
> it runs 24/7: no sleep, no distractions, no burnout
> it can actually learn skills and work through tasks while you're busy elsewhere
> it hands you leverage - more done in the same day, without hiring people

humans are capped: sleep, workouts, family, travel, attention. an agent running around the clock lifts that ceiling - and the work you never had hands for starts getting done on its own while you live your life.

Follow me to learn more about AI.

[billylwy22 @billylwy22]

Gate @Gate_zh 的 U 卡真的很值得辦，申請流程簡單，而且很多地區幾步就能免費拿到，用起來也很順手。

我自己會比較推薦這類卡的原因很直接：平時線上線下消費都能直接刷，能綁 Apple Pay 和 Google Pay，而且還有最高 5% 的 USDT / GT 返現。對經常需要海內外消費、又不想折騰太多支付流程的人來說，體驗會輕鬆很多。

申請方式也很清楚，基本就是：
Gate App → Pay → Gate Card → 選擇 Standard Card → 上傳資料 + 人臉辨識 → 等審核通過後就能直接使用虛擬卡。

地址說明也要看清楚：

馬來西亞、菲律賓、中國台灣、越南、泰國可直接用身份證申請，無需地址證明。

韓國可使用居民登記證申請，同樣無需地址證明。

我覺得它最方便的一點，是免費申請、沒有月費也沒有年費，拿來日常消費和綁定行動支付都挺實用。現在很多線上線下場景都能直接刷，整體絲滑度不錯。

申請連結：
https://t.co/InegP6ZYX5

[The Wolf Of All Streets @scottmelker]

This is why I believe in Bitcoin.

If you only watch 9 minutes of my content, ever, please let it be this.

[Solana @solana]

Jupiter COO @kashdhanda on the unique thing about Solana:

"Solana is the network that keeps getting better, and it's one of the few networks that actually does that. There's a focus on practical value and what actually works."

"A lot of the conversations are, 'How do we increase bandwidth? How do we reduce latency?' Those are the things that really matter."

"Applications on Solana do the majority of application revenue across any ecosystem, because the focus is on supporting builders, supporting companies and supporting real businesses."

[Knzo @0xKnzo]

CHINESE DEV JUST WIRED A FIBER-OPTIC HOME RACK AND MADE $290K LAST YEAR WHILE HIS EX-COFOUNDERS BURNED SERIES A ON ANTHROPIC BILLS

Silver rack with handles on top. Blue fiber loops into green LC connectors. Four HP mini-PCs stacked below.

Build cost $2,400. Power $45/mo. Four clients pay $6,000 each per month.

Pause at 0:04 — the LCD reads 192.168.4.2. Anthropic just billed his ex-team more than his entire rack costs.

Anthropic bills enterprises $24K/month for API access. He runs the same workflow on a $2,400 home fiber rack. Break-even in three days.

Fiber-to-the-bedroom just became AI's most under-priced asset. Anthropic isn't watching.
Full setup in the video.

[Chloe.eth ☕️✨ @ivanlp93]

serious question: you actually rebalance every quarter to lock gains or just hold bags hoping for the next leg up? trimmed my defi positions last month after decent run, kept the side income flowing without the usual drawdown stress how often you pull profits in practice?

[Chloe.eth ☕️✨ @ivanlp93]

now my nft side hustles run on autopilot. 🔥

[rvaniaaa @rvaniaaaa]

This guy turned Claude into a self-fixing AI developer. And it doesn’t stop after one try.

Most AI coding agents write the code once. If it has bugs, they hand it back to you. This is different. The AI keeps fixing its own work until everything passes.

It starts with two specialized agents. The builder writes the code. The checker runs the tests, type checks, and linting.

They never share the same job. The builder never reviews its own work. The checker never writes code. It only reports what failed.

An orchestrator connects them into a loop. Builder writes the feature. Checker runs every validation.

If everything passes, the loop ends. If something fails, the builder doesn’t give up. It gets the failures back, fixes them, and hands the code to the checker again.

This repeats automatically until the code is clean. But one decision keeps the whole loop reliable.

The loop has strict stop rules. It can only run five times.

And if a new fix breaks something that was already passing, the loop stops immediately. Otherwise the agents can loop forever or slowly weaken the tests just to fake a pass.

This is how you stop treating AI like a coding assistant and start treating it like an engineering team.

Bookmark this.

[Chloe.eth ☕️✨ @ivanlp93]

this codes my nft drops while i explore bali.

[rvaniaaa @rvaniaaaa]

This guy turned Claude into a self-fixing AI developer. And it doesn’t stop after one try.

Most AI coding agents write the code once. If it has bugs, they hand it back to you. This is different. The AI keeps fixing its own work until everything passes.

It starts with two specialized agents. The builder writes the code. The checker runs the tests, type checks, and linting.

They never share the same job. The builder never reviews its own work. The checker never writes code. It only reports what failed.

An orchestrator connects them into a loop. Builder writes the feature. Checker runs every validation.

If everything passes, the loop ends. If something fails, the builder doesn’t give up. It gets the failures back, fixes them, and hands the code to the checker again.

This repeats automatically until the code is clean. But one decision keeps the whole loop reliable.

The loop has strict stop rules. It can only run five times.

And if a new fix breaks something that was already passing, the loop stops immediately. Otherwise the agents can loop forever or slowly weaken the tests just to fake a pass.

This is how you stop treating AI like a coding assistant and start treating it like an engineering team.

Bookmark this.