Jibbie George

[Jibbie George @JibbieGeorge]

Early morning skiing is a whole vibe (Lực V. Triệu)

[Jibbie George @JibbieGeorge]

A good plate of Chinese dim sum can fix any bad day (Hỷ H. Đỗ)

[شيماء @chayma65589]

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

I❤️
محمد ﷺ
MUHAMMAD

[Eileen's mother ❣🍉 @rola_almoghrabi]

I beg you, just share😭😭
︎

︎

︎

︎

︎
︎

︎

︎

︎

we’ll remind you
︎

︎

︎

︎

︎
︎

︎

︎

︎

︎
︎

︎
︎

︎
︎

︎

︎

︎

︎

︎

︎

︎
that
︎

︎

︎

︎

︎

︎
︎

︎

︎

︎

︎

I desperately need milk for my daughter . €10
https://t.co/FLk3iM7URh

[Jibbie George @JibbieGeorge]

Prague at night is a whole different world (Tín T. Cao)

[Rich @RrichPRMR]

She's an SMM manager who makes $11,000 a month and half of it comes from filming herself do the exact job she gets paid to do for other people.

Sit with how backwards that is.

Her job is running other brands' social media.

So she runs her own and it became the highest-paid client she has.

Every video of her working is, quietly, an ad for hiring her.

Here's the loop most people in this field never figure out: her account isn't a distraction from the work, it's the proof of the work.

A potential client doesn't need her to explain that she's good at social media.

They just watched her grow her own to millions of views. The pitch and the portfolio are the same thing — she closes clients who found her by watching her be good at the job, no cold outreach, no proposals.

And that changes the whole income.

She lands agency clients at premium rates because they've already seen results before the first call.

She sells templates, presets, and mini-courses to the thousands of people who want to do what she does — built once, sold nightly.

And brands pay her to feature their tools mid-workflow, on top of platform payouts from millions of views.

Four income streams, all fed by one account that runs itself.

Because she barely touches the editing.

She films her work, hands the raw clips to AI, it cuts the strongest few seconds, captions, schedules the week.

The system that makes her content is the same system she sells — she's living proof it works.

The reframe for anyone with a skill: the most convincing ad for what you do is footage of you doing it.

Don't tell people you're good.

Post the work until the work speaks.

Whatever you'd pitch a client, film it instead — the clip closes better than the pitch ever will.

What's a skill you could sell just by filming yourself doing it well?

Drop it below — best ones get their own breakdown.

[Jibbie George @JibbieGeorge]

Prague at night is a whole different world (Tín T. Cao)

[Jibbie George @JibbieGeorge]

Someday I'll make it back to Porto (Minh Đ. Chu)

[dr @Rtiji3]

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

♥️ I LOVE ALLAH♥️

[Lana & Hala Family's 🇵🇸🍉 @hazemkhilla]

MY CHILD IS DYING, PLEASE😭

︎
︎

︎

︎

︎
︎
︎

︎

︎

︎

︎don’t forget me🥺
︎

︎

︎

︎

︎

︎
︎

︎
please care

︎
︎

︎
︎

︎
︎

︎

︎

︎

︎
︎

︎

I beg you, I need treatment.

I kiss your hands

Help my sick child, he's dying

https://t.co/ZA6ngSZBuj

[Ibrahim Kamel @abrahimKamal74]

Unfortunately, no one cares 😭

︎
︎

︎

︎

︎

︎
︎

︎

︎
︎

we’ll remind you
︎

︎

︎

︎

︎
︎

︎

︎

︎

︎
︎

︎
︎

︎
︎

︎

︎

︎
︎

︎
that
︎

︎

︎

︎

︎

︎
︎

︎

︎

︎

I desperately need milk for my baby😭
https://t.co/zEGGLAZtW8

[KyzoroX @kyzoroXX]

The GPUs weren't the problem.

245kW was.

The engineers spent months comparing AI chips.

MI355X vs B300.

Memory.

Bandwidth.

Cost.

Performance.

The purchase was approved.

Then reality hit.

One fully loaded AMD Helios rack would draw up to 245kW.

Suddenly nobody was talking about GPUs anymore.

They were talking about pumps.

Pipes.

Heat exchangers.

Water flow.

Backup cooling.

Because Helios isn't just another AI server.

It packs:

• 72 Instinct MI355X GPUs
• 31TB of HBM memory
• 2.9 ExaFLOPS (FP4)
• 1.7PB/s memory bandwidth

Air cooling simply isn't enough.

Every GPU, CPU and NIC sits on a direct liquid-cooling loop.

The biggest challenge in AI hardware isn't building faster chips anymore.

It's finding a way to remove all that heat.

[CBS News @CBSNews]

The battle against France's largest wildfire since World War II isn't over.

Fire officials say underground hotspots are still burning beneath the forest floor and could reignite days or even weeks later.

[Defn Pak @DefnPak]

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah

Allah
♥️ I LOVE ALLAH♥️

[Jibbie George @JibbieGeorge]

Something about orange light makes everything feel creative (Tuấn V. Chu)

[Jibbie George @JibbieGeorge]

A gray day with a still energy to match - Chước T. Đoàn

[nOnEpcbl. @nonepcbl]

everyone is asking whether kimi k3 can beat claude

the useful question is how much of your claude bill never required claude-level intelligence

whole repos. research archives. second brains. agent loops rereading the same files

these jobs don't need the smartest model every time

they need a model that can afford to look at everything

k3 gives them a 1m-token window at a flat per-token rate

$3 per million input tokens

$0.30 when the stable context hits cache

same files. same model. 10x cheaper rereads

but max reasoning is always on at launch, so using it for quick edits defeats the point

keep the hardest reasoning and security-sensitive work on frontier models

route the massive, repetitive context jobs to the volume machine

k3 doesn't need to be the smartest model in your stack

it only needs to stop your smartest model from doing the expensive boring work

[Vettan @Vettan0]

THIS FOUNDER JUST WIRED HIS ENTIRE DIGITAL CONSCIOUSNESS INTO A SELF-IMPROVING AI LOOP.

Your Second Brain is probably just a graveyard of dead links and forgotten Notion pages. You save articles you never read and write ideas you instantly lose.

Human memory does not scale, and manual organization is a massive waste of time.

This system fixes the bottleneck by turning a passive database into an active, autonomous cognitive engine. He integrated Tiago Forte’s PARA organization with Andrej Karpathy’s LLM Wiki concept—but removed the human element entirely.

By giving OpenClaw terminal access, the AI acts as a 24/7 librarian.

Every single day, the agent silently sweeps his local files. It captures his random thoughts, the articles he reads, and the projects he finishes. Without any manual tagging, Claude categorizes the raw data, maps the hidden connections, and distills it into a massive, interconnected knowledge graph.

When he sits down to work, the context is already built.

The ultimate leverage is the flywheel effect. The moment he creates and publishes a new project, the AI ingests that final output right back into the system. The more he works, the denser the graph becomes, and the smarter the AI gets at replicating his exact thought process.

He stopped managing his files and started compounding his intelligence.

[Jibbie George @JibbieGeorge]

That feeling when someone texts you exactly when you were thinking about them - Dâu T. Vương

[KyzoroX @kyzoroXX]

Four Intel Arc Pro B70s. 128GB of VRAM total. $4,000.

One RTX PRO 6000 Blackwell. 96GB. $10,000.

The pitch writes itself: gang up cheap cards, beat the expensive one, pocket $6,000. And on raw memory it's true — four B70s hold more than the single Blackwell.

Here's what the price comparison leaves out. Four cards means the model gets split across four of them, and every token crosses PCIe to move between cards. The Blackwell is one pool — no splitting, no PCIe tax. Same reason a 128GB cluster and 128GB unified aren't the same 128GB.

So the real question isn't "which has more VRAM for less." It's what you're running. Models that fit on one B70's 32GB? The cluster's a steal. Models that need to span all four? You're now paying in latency what you saved in cash.

$4,000 of Intel is the right call for a lot of workloads. Just not because it "beats" a $10k card — because it's a different shape of machine for a different job.