YOG
516 posts
YOG
@Yogabool
Ancient wisdom for modern business. The greatest teacher, failure is. AI Business • Startups • Macro
Inscrit le Mayıs 2026
179 Abonnements103 Abonnés
A 20-year-old guy from Japan created an AI construction girl and earned $3,737 in 2 weeks
He used Claude to create photos and videos
He came up with a new niche, "cute construction girl" and started a TikTok for her, which already had 54,328 new followers after 4 days
Men started writing to her in private messages non-stop: they liked her a lot, and he decided to start a Fanvue
TikTok is for brand awareness, and with the help of Fanvue, this awareness turns into money
In 2 weeks, 624 men signed up for a paid subscription for $5.99. There, the girl does construction work, but without clothes on
A little imagination + AI and your possibilities are unlimited
Bober_smart@Bober_smart
English
ANTHROPIC'S $2.4M LEAD ENGINEER LEAKED THE INTERNAL OBSIDIAN BRAIN AND WAS FIRED THE SAME DAY
A private vault mapping the full model architecture. 22 neurons per layer, ReLU activations, weighted paths across every hidden layer.
Feature inputs labeled, forward-propagation exposed, training notes attached.
That is already the wrong question to ask about loyalty.
The vault was not source code.
It was the reasoning map behind how the model is shaped, layer by layer, decision by decision.
The kind of document competitors pay nine figures to reconstruct from scratch.
Here is where it stings.
One engineer, one export, one afternoon.
Four years of internal alignment work sitting in a Markdown folder.
The public sees a leak.
The industry sees a blueprint.
Hex Horizon@Noderunner_Hex
English
THIS GIRL BUILT A SECOND BRAIN IN 2 WEEKS WITH CLAUDE + OBSIDIAN
The interesting part is not the graph view.
The graph is just the receipt.
The real system is simple:
Every book note goes in.
Every meeting note goes in.
Every email worth remembering goes in.
Every video, idea, and half-formed thought goes in.
Obsidian becomes the place where the memory lives.
Claude becomes the assistant that helps turn messy input into structured notes, links, summaries, missing connections, and better questions.
That is why this works better than a normal notes app.
Most people use notes as storage.
They write something down, forget the file name, then search for it 3 months later.
A second brain only becomes useful when notes start talking to each other.
A note about dopamine can connect to ADHD.
ADHD can connect to productivity.
Productivity can connect to meeting habits.
Meeting habits can connect to personal branding, work output, or content ideas.
That is the compounding loop.
You are not just saving information.
You are building a map of what you already know, what you keep repeating, and what you still do not understand.
The underrated part:
Claude is not the brain.
If you let AI generate everything from scratch, you get polished noise.
The useful workflow is using Claude on top of your own inputs:
clean this note
extract the key claims
link it to related notes
find contradictions
show missing context
turn this into next actions
That is a real advantage.
Not because it makes you smarter overnight.
Because it makes your thinking searchable, connected, and harder to lose.
The caveat is obvious:
A second brain can become a graveyard if you only collect notes.
The output has to be decisions, writing, projects, or better questions.
Otherwise it is just a beautiful Obsidian graph with no real use.
kocer@kocer_eth
English
A 22-YEAR-OLD BUILT A $42K/MONTH AI FILM STUDIO FOR $60.
Hollywood spent $200M to keep you watching.
He spent $60 to replace the system.
Two decisions:
• Pick the story
• Pick the style
Everything else is automated.
Claude + AI tools handle it all:
Script. Characters. Dialogue. Voice. Animation.
This week’s episode:
A cat in a chef’s hat cooking like a human while his boss yells from off-screen to hurry up with the sauce.
The cat keeps cooking.
No reaction. No panic. Just tension building in real time.
The pan sizzles. The argument escalates. The scene feels alive.
Two episodes drop every week.
Fully automated.
While he sleeps.
Revenue stack:
• Ads
• Patreon
• Sponsors
• Premium revenue
• Brand deals
Last month: $42,000 profit.
Tool cost: $140.
One sponsor paid $9,800 for a single episode.
Hollywood builds armies of people.
This builds content like a machine.
And somehow, a cooking cat ignoring chaos off-screen beats most “serious” productions.
English
She filmed her transformation every day for 8 months.
It made her $38,000.
The weird part? She barely edited a single video herself.
Day one, she wasn't a creator.
Just a girl starting over, who pointed a phone at it.
Every session.
Every tired face after the last set.
Every small win.
No scripts, no polish.
Here's what everyone underestimates: a transformation filmed in real time is the most bankable content in fitness.
Not because it teaches anything.
A workout video gets a view.
A story gets a following.
People come back tomorrow to see what happens next.
Her daily cost was minutes.
She filmed raw clips at the gym — AI did the rest.
The cuts, the captions, the on-screen text, the posting schedule.
240+ daily posts in 8 months, and she never opened an editor.
Alone, that volume is impossible.
With AI, it's a habit.
Growth looked like every transformation: invisible for two months, then all at once.
By day 140, she wasn't pitching brands anymore.
Brands were pitching her.
Because an ad inside "day 142 of my transformation" doesn't feel like an ad — it feels like part of a story people are rooting for.
Brands pay a premium for that.
She picked who got in, set the terms, and stacked sponsored posts, monthly partnerships, and affiliate links underneath.
Total across 8 months: about $38,000.
From footage of a process she was living anyway.
Costs: a phone and ~$40 a month in tools.
Now the part that should sting.
Every day she trained, millions of people did the exact same thing.
Same gyms.
Same effort.
Same sweat.
The only difference? She pressed Record.
Most people finish a transformation with a few before-and-after photos.
She finished hers with an audience, recurring brand deals, and $38,000.
Same journey. Different decision.
Full playbook below👇
Rich@RrichPRMR
English
She films 15-second gym clips and makes $7,700/month. Not from brand deals. From a $34 program AI helped her build. Here's how.
Her clips do one job: make you stop scrolling.
No teaching, no talking.
Fifteen seconds of a clean set, and every clip ends the same way — free starter guide in bio.
That guide is the trap, in the good sense.
To get it, you drop your email.
A follower is rented from the algorithm.
An email is hers forever and a list buys at a completely different rate than a feed.
The system, step by step:
She trains 4 times a week and films 2–3 exercises each session.
AI cuts that into ~20 clips, writes the hooks, and schedules the week across TikTok, Reels and Shorts.
Her editing time: zero.
The clips pull 2-3M blended views a month.
About 2,500 of those viewers grab the free guide.
Her list grows by that much every month, compounding.
Once a month she emails the list one offer: a $34 twelve-week program.
She wrote the training herself — AI structured it into weeks, wrote the explanations, formatted the PDF.
Built once, updated never.
Roughly 1 in 11 new subscribers buys, plus a steady drip from older subscribers.
That's ~225 sales a month. 225 × $34 = $7,650.
Round it with a few affiliate sales from the guide itself — $7,700.
Costs: about $40 in tools. No inventory, no clients, no calls.
The girl filming next to her posts the same set, gets 40 likes, starts over tomorrow. Same footage. One collects likes. The other collects emails.
Attention is free.
She just refuses to waste it.
Full playbook below👇
Rich@RrichPRMR
English
THIS GUY SHOWED HIS DASHBOARD AND NOBODY REALIZED THEY WERE STARING AT $24,000 A MONTH
Leads tracked. Pipeline staged. Revenue updating in real time. All of it running on autopilot while he slept.
He built it for a marketing agency. Six people doing ops by hand. $32,000 a month in salaries. Five agents replaced 80% of it for $180 a month. He charges $3,000 to keep the whole thing humming.
They watched a lead come in. Get scored. Get a reply in 60 seconds. Land in the CRM. Nobody touched a thing. They signed that same week.
They never cancelled. Cancelling means hiring the team back. That math does not math.
10 clients now. $24,000 a month. The dashboard updates itself. He just watches the numbers climb.
Bookmark this. The agency model just got rewritten by one guy and a quiet dashboard.
English
🇺🇸 U.S. Secretary of State Marco Rubio says the U.S. government plans to introduce facial recognition for passport applications within the next few months, allowing applicants to verify their identity using a photo instead of visiting places like Walgreens or CVS for a passport picture.
According to Rubio, the goal is to simplify the process and save applicants time.
"Our security system will verify the facial ID," he said.
English
🤡 Trump said that whenever his kids invest in stocks or do anything like that, people think they have insider information.
I kinda feel bad for my kids.
No matter what they do, even if they're just buying a truck, people assume they've got insider info.
If they buy a cupcake company, people somehow tie the energy used to bake those cupcakes back to my energy policy.
So they're basically in a constant conflict-of-interest situation.
English
YOG retweeté
🇺🇸 America's 250th birthday celebration is already lighting up the sky.
A stunning air show featuring 2,500 drones transformed the night over North Richland Hills, Texas, with massive images of a rocket, a bald eagle, and George Washington.
The patriotic display, created by Sky Elements Drone Shows, gave local families an early look at the celebrations leading up to the 250th anniversary of the signing of the Declaration of Independence.
Gunther Eagleman™@GuntherEagleman
🚨 TEXAS DOES AMERICA 250 RIGHT, 2,500 DRONES LIGHT UP THE SKY OVER DFW! Everything’s bigger in Texas, including Fourth of July celebrations. North Richland Hills put on an epic drone show with 2,500 drones forming Uncle Sam, a massive red, white, and blue eagle, plus pyrotechnics for extra flair. Iconic American imagery lighting up the night. This is how you honor 250 years of freedom. Texas knows how to celebrate the greatest country on Earth! God bless Texas and God bless America!
English
YOG retweeté
THIS MODEL JUST MATCHED GPT 5.5 AND OPUS 4.8 ON REAL PHYSICS FOR ZERO DOLLARS. ZERO.
Same prompt. Three HTML5 canvas scenes. Real physics. Four models. One of them charged nothing and still won.
The test: a cannon demolishing a brick wall, a bowling ball knocking down pins, a tornado sucking in random objects. Real physics. No shortcuts. No excuses.
LongCat: 18,015 tokens, $0.00
Opus 4.8: 18,872 tokens, $0.48
GPT 5.5: 32,588 tokens, $0.98
GLM 5.2: 31,062 tokens, $0.09
LongCat beat Opus 4.8 and GLM 5.2 on physics. Cleaner collisions. Nothing clipped. Nothing fell through the floor. On detail and rendering it matched GPT 5.5, the best looking output of all four. And it billed exactly nothing.
The model that cost zero dollars just smoked models that cost half a dollar and nearly a dollar. The price to performance curve just snapped in half.
Bookmark this. Free just matched premium and the premium models are running out of places to hide.
100F.exe@100F_exe
🔥 FABLE 5 JUST DOMINATED THE LATEST PHYSICS TEST AND THEN HANDED THE BILL. IT COST 6X MORE THAN OPUS AND 39X MORE THAN GLM. Same prompt. Three HTML5 canvas scenes. Real physics. Four models. One winner. One very awkward invoice. The test: car into a brick wall, wrecking ball through a house, catapult launching a rock at a castle wall. Real physics. No shortcuts. Fable 5 won every scene. Cleanest physics. Best overall results. But here is what it cost: Fable 5: 62,158 tokens ($3.12) GPT 5.5: 37,753 tokens ($1.14) Opus 4.8: 22,280 tokens ($0.56) GLM 5.2: 36,246 tokens ($0.08) Fable 5 took the crown. GPT 5.5 came the closest and even edged it out slightly in the monster truck demo. GLM 5.2 won zero scenes but charged pocket change. Eight cents. Total. The benchmark ran on atomic dot chat, a desktop app that runs LLMs locally. No cloud markup. Just raw token cost. The performance gap is real. The price gap is absurd. The king won but the king charged 39 times more than the model that cost eight cents. Bookmark this. The next battle is not about who wins. It is about who wins for a price that does not feel like a car payment.
English
YOG retweeté
The real AI revolution isn't about replacing jobs. It's about redefining the relationship between humans and work. If machines create most of the world's value, then our biggest challenge won't be productivity - it will be purpose....
100F.exe@100F_exe
ELON JUST KILLED WORK. NOT THE 9 TO 5. THE ENTIRE CONCEPT. AI and robots will do everything. Every job. Every task. Every skill you spent decades sharpening. Gone. Handled by a machine that never sleeps, never complains, never asks for a raise. And the result? Universal high income. Not universal basic income scraping the bottom of survival. High income. You wake up. The machines have already run the economy for the night. Your money is there. Your time is yours. Work becomes a hobby. A choice. Something you do because it amuses you, not because rent is due. The factories are already silent except for the hum of robots. The agents are already writing code, editing video, closing deals. The businesses are already running themselves while their owners sleep. This is not a prediction. It is a live feed. The real mindfuck nobody is ready for: when work is optional, who are you. When the thing you did for 40 years is done better by a machine that does not blink, what do you do with your hands. What do you do with your brain. What do you do with meaning. Bookmark this. Work just became a vintage concept. And Elon said it out loud like it was Tuesday.
English