Pinned Tweet
This is f*cking insane. Elon Musk in 2008, in a living room, 12 days from bankruptcy:
"We're going to make it
Or we're going to die trying"
In 4 minutes, he pitches a $109,000 electric car to a room of strangers while his company is almost out of money
12 Days From Bankruptcy → Living Room Pitch → Tesla → The Richest Man on Earth
Today Tesla is teasing FSD v15. It started in a living room
Bookmark and watch it whenever you feel like quitting
Then read how to build something that runs while you sleep: Harness, Loops, Graphs ↓
This is f*cking insane. Anthropic just went public at $1.5 trillion. Palantir CEO Alex Karp on Dario and his crew:
"They want to tell you we have to march into a future where we own nothing
Where none of us have jobs"
In 10 minutes, Karp calls the frontier labs parasites and ends the interview when Sam Altman comes up
Frontier Labs → All the IP → $1.5 Trillion → A Future Without Jobs?
Billionaires are fighting over who owns the future. Make sure you own a piece of it
Bookmark and watch it today
Then read how to build your own piece: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
30
671
This is f*cking insane. Turing Award winner Yann LeCun:
"Basing agentic systems on LLMs is a recipe for disaster
How can a system plan a sequence of actions if it can't predict the consequences?"
In 34 seconds, he explains why most agents fail: they act before they understand what their actions will do
Act → No Prediction → Wrong Plan → Broken Agent
Agents only work when there's a loop that checks every step
Bookmark and watch it today
Then read how to build agents that check themselves: Harness, Loops, Graphs ↓
This is f*cking insane. Andrej Karpathy just said your kids should ignore most of school:
"80% of education should be math, physics, CS"
In 17 minutes, he explains why: it carves grooves in the brain that get harder to carve the older you get
Pre-AGI World → It Gets You a Job → Post-AGI World → It Keeps You Empowered
Everything else you can tack on later
Bookmark and watch it today
Then read what to learn first: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
1
41
2,027
This is f*cking insane. Billionaire Mark Cuban on the next job wave:
"Software is dead because everything's gonna be customized to your unique utilization
Who's gonna do it for them"
In under 3 minutes, he explains who gets hired next
Not prompt engineers. Not people who use AI like a better Google. The ones who can walk into a company and make AI actually do the work
Software → Custom Software → Custom AI → The People Who Build It
Bookmark and watch it today
Then read how to become one of them: Harness, Loops, Graphs ↓
This is f*cking insane. Jensen Huang, the man who built Nvidia:
"Nobody writes prompts anymore
The new job is to write and handle loops"
In 51 minutes, he explains the shift that's going to define the rest of 2026
Prompts → Agents → Loops → Loops Running Loops → The New Job
He's selling the chips. He knows exactly what people are running on them
Worth more than any $2,000 AI course
Bookmark and watch it today
Then read how to write your first loop: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
40
651
This is f*cking insane. SpaceX just caught Starship with the tower. Elon Musk said this a year ago:
"If there are historians in the future, they will look back at Starship and say it was one of the most profound things that ever happened
The degree to which it is revolutionary is not well understood"
In under 3 minutes, he explains why full reusability changes everything
Rocket → Reusable Rocket → Starship → Mars → A Multiplanetary Species
AI is on the same curve. The people building on it now are the ones historians will mention
Bookmark and watch it today
Then read how to start building on it: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
1
44
3,232
This is f*cking insane. Fable 5.2 learned all of music theory and wrote a house track it can't even hear:
"It tries thousands of random patterns for every instrument, keeps the best one and arranges it into a 72-bar track"
Sound on. In 6 minutes you hear what a model does when you give it a loop instead of a prompt
Music Theory → 1000s of Patterns → Score Each One → Keep the Best → A Finished Track
Same trick works for code, ads and research. Generate, judge, keep, repeat
Bookmark and watch it today
Then read how that generate and judge loop is built: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
2
39
769
This is f*cking insane. Boris Cherny, the man who built Claude Code:
"Going from agents to loops is as big a jump as going from code to agents"
In 41 minutes, he shows exactly how he works now
Loops run his code reviews, his maintenance and his fixes, on their own
Code → Agents → Loops → Work That Happens Without You
Most people haven't made the first jump yet
Worth more than most $1,500 agent courses
Bookmark and watch it today
Then read the full breakdown of loops: Harness, Loops, Graphs ↓
This is f*cking insane. Cursor CEO Michael Truell on how 4 MIT kids built the fastest growing software company in history:
"We were four 20-somethings running the whole service with 5 people, and we were already a double-digit share of the AI labs' API revenue"
In this 28-minute talk, he walks through it with zero marketing and a first version built in 2 weeks
Failed CAD Startup → AI Editor From Scratch → $2B a Year → A $60 Billion Offer From SpaceX
Most people still write every line of code by hand
Worth more than any $20,000 startup accelerator
Bookmark and watch it today
Then read how small teams run agents now: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
37
2,120
This is f*cking insane. All-In's David Sacks on why Salesforce just let Claude sit on top of its own product:
"He's willing to allow Claude to be the front end, to have the primary user relationship
You don't want to go into every SaaS app to write your agents. You want to create them in the AI you use and run them across all your platforms"
In 2 minutes, he explains why the biggest CRM on the planet agreed to become a database that Claude agents read and write
Database → Apps → Agents → Claude On Top Of Everything
The app you log into every morning is turning into a storage box
Bookmark and watch it today
Then read how to run agents across all of it: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
1
34
559
This is f*cking insane. Anthropic engineer:
"Prompting Fable 5.2 is a waste of the model
95% of our engineers already run self-improving loops on it. The next move is graphs"
In 10 minutes, she builds her whole Claude Code setup live, starting from an empty terminal
Empty Terminal → Agents → Loops → Graphs → Self-Improving System
Prompting was the old way. Engineering the system around the model is the new one
Worth more than most $1,000 agent engineering courses
Bookmark and watch it today
Then read why your agent is not autonomous yet, and the loop that fixes it: Harness, Loops, Graphs ↓
This is f*cking insane. Jane Street engineer:
"95% of our quants run swarms of AI agents
We built our own AI in 3 days, and now it does research and trading for a firm that makes $39.6B a year"
In 78 minutes, the Jane Street team shows how AI lives inside one of the most secretive funds on Earth
Research Agents → Big Data → Own Data Centers → AI That Watches Live Trading
Wall Street isn't using AI as a chatbot. It's running it as a team
Worth more than any $5,000 trading course
Bookmark and watch it today
Then read the loop these teams run so their agents check themselves: Harness, Loops, Graphs ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
3
35
746
This is f*cking insane. Anthropic engineer:
"You're not supposed to prompt Opus 5.5. You're supposed to build the system that prompts it for you"
In 33 minutes, she shows how the team behind Claude builds agents that remember, catch their own mistakes and get sharper every run
Prompt → Harness → Memory → Self-Check → An Agent That Improves Itself
Most people are still rewriting the same prompt. The people who build the model quit doing that a year ago
Worth more than any $2,500 agent course
Bookmark and watch it today
Then read why your agent is not autonomous yet, and the loop that fixes it ↓
This is f*cking insane. Microsoft CEO Satya Nadella:
"Microsoft has 200,000 employees. Soon we'll have 20 million agents, all running in loops
You give it a high-level intent and it goes off and does the whole job"
In this 1-hour talk with LinkedIn founder Reid Hoffman, he explains why the agent stopped being a tool and became the company
Autocomplete → Chat → Agent Mode → Fire And Forget → 100 Agents Per Person
He already runs 100 agent sessions at once. Microsoft had to build a new app just to manage them
Worth more than most $3,000 AI courses
Bookmark and watch it tonight
Then read the motion engineering article: one AI hire, one job, nothing ships without a yes ↓
I fixed the same bug in two films
Opus never remembers my fixes
So I hired an AI employee to remember them for me, then tried to make him break my one rule on deadline
He didn't ↓
1
37
563
This is f*cking insane. OpenAI CEO Sam Altman to a room of Stanford students:
"You don't need to write prompts anymore. Tell it what you want done, then let it work"
In 38 minutes, he shows how to use AI better than 99% of people do
Prompts → Context → Agents → Systems That Prompt Themselves
Most people are still using about 15% of what it can do
Worth more than any $1,000 AI course
Bookmark and watch it tonight
Then read the motion engineering article: 100+ defects across 10 Opus 5.5 films, and the AI hire that catches them ↓
Sasha King retweeted
This is f*cking insane. Satya Nadella spent a whole day on coding interviews at Microsoft. Then came the last question:
"You're crossing a street, a child falls. What do you do?"
He said "call 911". The interviewer told him: "You pick the child up and hug them"
Today he runs Microsoft. Machines can code now. That's the part they can't learn
Learn the rest for free:
Andrew Ng's Agentic AI course
deeplearning.ai/courses/agen…
Anthropic Academy
academy.claude.com/
Claude Code 101
anthropic.skilljar.com/claud…
Anthropic's prompt engineering course
github.com/anthropics/course…
Hugging Face agents course
huggingface.co/learn/agents-…
Hugging Face LLM course
huggingface.co/learn/llm-cou…
Karpathy's Neural Networks: Zero to Hero
karpathy.ai/zero-to-hero.htm…
Practical deep learning
course.fast.ai/
Google's machine learning crash course
developers.google.com/machin…
AI for Everyone
coursera.org/learn/ai-for-ev…
All of it costs $0
Bookmark it
Then read the "Harness, Loops, Graphs" guide below ↓
This is f*cking insane. Satya Nadella spent a whole day on coding interviews at Microsoft. Then came the last question:
"You're crossing a street, a child falls. What do you do?"
He said "call 911". The interviewer told him: "You pick the child up and hug them"
Today he runs Microsoft. Machines can code now. That's the part they can't learn
Learn the rest for free:
Andrew Ng's Agentic AI course
deeplearning.ai/courses/agen…
Anthropic Academy
academy.claude.com/
Claude Code 101
anthropic.skilljar.com/claud…
Anthropic's prompt engineering course
github.com/anthropics/course…
Hugging Face agents course
huggingface.co/learn/agents-…
Hugging Face LLM course
huggingface.co/learn/llm-cou…
Karpathy's Neural Networks: Zero to Hero
karpathy.ai/zero-to-hero.htm…
Practical deep learning
course.fast.ai/
Google's machine learning crash course
developers.google.com/machin…
AI for Everyone
coursera.org/learn/ai-for-ev…
All of it costs $0
Bookmark it
Then read the "Harness, Loops, Graphs" guide below ↓
This is f*cking insane. Mark Zuckerberg and Alexandr Wang sat down at a pub for 2 hours and explained how Meta is betting $100 billion on AI
How frontier models get built, why AI won't be winner-take-all, and what it's like working for Zuckerberg
Muse Wonder → $100B Bet → Agentic Commerce → Beyond the Smartphone → Superintelligence Labs
The most open look at Meta's AI plan ever filmed
Worth more than any MBA
Bookmark it and watch it this weekend
Then read the "Harness, Loops, Graphs" guide below ↓
Your agent is not autonomous
You are its loop
You check every step because nothing else in the system can
I broke down the 7 levels from babysitting agents to running a fleet ↓
42
824
Sasha King retweeted
This is f*cking insane. Andrew Ng just recorded the best 2 hours on graph engineering I've ever seen
"Your agents don't need a bigger model. They need a graph"
He goes from your first working agent to a system that runs the whole thing for you
First Agent → Loops → Graphs → Agents That Improve Their Own Code → A System That Runs Itself
Anyone can build one agent. Almost nobody builds the graph
I've seen $500 courses that teach less than this video alone
Bookmark and watch it today
Then read the "Harness, Loops, Graphs" guide below ↓



