Karmi

Ai researcher / YOUR money is just YOUR money

[Karmi @0xKarmi]

https://t.co/f4YJLEmGxO

[Karmi @0xKarmi]

The Starbucks CEO, under oath, on his own money: "I earned it, no one gave it to me."

40,000 stores. Its founder grew up in public housing and said so under oath.

He grew up in public housing in Brooklyn. His parents never owned a home.

He was sitting in front of a Senate committee being asked why a company that size fights its own workers when they organise, and that was the sentence he chose.

Both halves are true and they sit badly together. The distance he travelled is real. So is the fact that he was being asked about the people currently standing where he started.

The story that got him there is also the story he uses to answer for where he ended up.

That is what makes it the most honest moment in the hearing, and the least comfortable one.

[Karmi @0xKarmi]

https://t.co/f4YJLEmGxO

[Karmi @0xKarmi]

Elon Musk told people SpaceX had maybe a 10% chance of working before they joined.

He says the number is going to be high. He says the odds are poor. He says it will probably fail.

Then people who stay have already agreed to the terms, and the ones who would have complained left on their own.

Nobody argues you down from an expectation you set at the ceiling.

Most people undersell the difficulty to get the yes, then spend the rest of the project defending it.

Article below 👇

[Karmi @0xKarmi]

https://t.co/f4YJLEmGxO

[Karmi @0xKarmi]

Warren Buffett: if you can't stomach a 50% drop, you are not ready for long-term wealth.

Berkshire's own stock has fallen roughly half three separate times while he was running it.

He did not sell. He did not hedge it. He said out loud, in advance, that it would happen again.

The people who compounded alongside him were not smarter. They were the ones who did nothing on the worst day.

Everyone thinks they are testing their strategy. They are testing whether they can watch a number go down.

[Karmi @0xKarmi]

https://t.co/f4YJLEmGxO

[Karmi @0xKarmi]

5,000,000 views on a free lecture by a Goldman Sachs vice chairman.

his advice to a room of students was not a framework. it was: read the Wall Street Journal every single day.

not skim it. not read the summary. every day, the whole topic, page by page, until the names in the industry stop being strangers.

his lectures on career strategy are on youtube and have over five million views. free. no funnel. a Goldman managing director explaining exactly what gets people promoted, to people who haven't started yet.

meanwhile there is a $2,000 "break into banking" program selling a compressed version of the same talk.

the daily habit is free and boring. the course is expensive and exciting. only one of them works.

bookmark & read article ↓

[Karmi @0xKarmi]

https://t.co/f4YJLEmGxO

[Karmi @0xKarmi]

@Ifeanyiuchend Your food looks delicious

[Karmi @0xKarmi]

a 22-year-old turned "cartoon characters as real people" into $10,000 a month. one format, run forever.

- the hook: a character everyone knows, rebuilt as a believable real person, revealed next to the original
- the engine: Midjourney for the realistic design, Runway for the motion, Claude to script the reveal, CapCut to stitch it, one clip an hour
- the fuel: nostalgia plus the "does it actually look like them" debate that drags every comment section into a fight
- the moat: he's not drawing, he's casting, so he can do any show, any character, on demand
- the money: brand deals, a paid preset pack, and creator payouts across every platform at once

he didn't animate a thing. he built a machine that turns childhood into scroll-stopping clips.

the full setup is in the article below.

[Ridark @ridark_eth]

https://t.co/nbKGZnMx2m

[Karmi @0xKarmi]

most people see a 25-year-old cartoon. he sees $11,000 a month.

he takes characters everyone grew up with and rebuilds them as photoreal humans, standing right next to their cartoon selves, same pose, same street. the split-screen is the whole hook.

- pick a character burned into everyone's memory
- Midjourney generates the realistic human version, matched to the original
- Runway walks them down the same sidewalk as the cartoon
- CapCut lays the two side by side and posts it

people don't scroll past their childhood standing next to a real person. the comments argue about the likeness, and the argument feeds the algorithm.

everyone else sees nostalgia. he saw a format that prints $11,000 a month. the build is in the article below.

[Ridark @ridark_eth]

https://t.co/nbKGZnMx2m

[Karmi @0xKarmi]

@ghostinlatent He looks like Baza from the cartoon

[Karmi @0xKarmi]

@Abobsterina To be honest, she has a big heart

[Karmi @0xKarmi]

@casper_calls Donald Trump is the best, as always

[Karmi @0xKarmi]

@Abobsterina She’s beautiful, and she has a big heart

[Karmi @0xKarmi]

@ghostinlatent This robot is having fun vacuuming

[Karmi @0xKarmi]

@veloraaivs Hahaha, it’s funny how they’re cutting Elon Musk’s hair.

[Karmi @0xKarmi]

@promptixAI And why do we need workers now when we have a robot?

[Karmi @0xKarmi]

@casper_calls Elon Musk has always been admired as a person

[Karmi @0xKarmi]

@ridark_eth I saw Edith, he's driving there in a $100,000+ car

[Karmi @0xKarmi]

@Yumzlef Jensen Huang is one of the greatest people in my opinion

[Karmi @0xKarmi]

Memory and continual learning: context → learned weights → an agent that knows what you did yesterday

"you don't have to write monstrous system prompts, they're already there"

here's the shift, step by step:

step 1 → 1:35 – understand the context, don't just re-read files at test time
step 2 → 4:24 – it works on any transformer – this isn't a special-model trick
step 3 → 4:41 – spend compute up front to learn your company's way of working
step 4 → 6:56 – you have to remember things to compose them into anything complex
step 5 → 26:26 – it just knows what you were working on yesterday

state isn't a bigger context window. it's what survives the window closing.

most people pay for the same context on every call and call it memory

watch & bookmark – then read Loop Engineering↓

[codila @0xCodila]

https://t.co/UJL7o2l9oR

[Karmi @0xKarmi]

@insaneposes My mom decided to wash my jeans one day, and there was $200 in there