1% to learn ai 99% to ...
heaven
Joined July 2026
- Tweets39
- Following33
- Followers93
- Likes286
Pinned Tweet
This is f*cking insane. Anthropic CEO Dario Amodei just told software engineers how long they have left:
"I actually think software might go even faster. We're already in our Centaur phase for software, and during that phase the demand for software engineers may go up
But the period may be very brief. This is happening over low single-digit numbers of years. How do we get people to adapt fast enough?"
In 2 minutes, the man building Claude explains why coding, not data entry, could be the first job AI finishes off
Copilot → Centaur → Supervisor → Just The Machine
After Deep Blue beat Kasparov, a human checking the AI beat any human and any machine alone for 15-20 years
Then that era ended, and now it's just the machine
He thinks software gets a few years of that phase, not 20
The engineers still standing after it will be the ones who stopped typing code and started running the agents that write it
Worth more than any $5,000 coding bootcamp
Bookmark it and watch it tonight
Then read how to go from writing code to running the agents that write it in the harness engineering guide below ↓
This is f*cking insane. Claude Code creator Boris Cherny:
"The biggest mistake people make with Claude Code - they keep feeding a new model the instructions they wrote for the old one
Every six months, delete your claude.md, delete your skills, delete your hooks. See what the model does. It might surprise you"
Opus 5 → Fewer Instructions → Leaner Harness → An Agent That Does More On Its Own
what Boris really means:
step 1 → every rule you wrote was a crutch for a weaker model. wipe them and test the new one bare
step 2 → he says it straight for Opus 5: the model might just not need all the instructions you needed for past models
step 3 → your setup isn't the product, the model is. keep only what it still gets wrong and let it do the rest
"if you're willing to throw your setup away every six months, you're already ahead. not the teams with 2,000-line rule files. you"
Watch it today before it gets buried
Then read how to build a harness that stays lean as the models get smarter in the guide below ↓
watch it now ↓ nitter.gq/angelicsmind/status/21…
This is f*cking insane. Anthropic CEO Dario Amodei just told software engineers how long they have left:
"I actually think software might go even faster. We're already in our Centaur phase for software, and during that phase the demand for software engineers may go up
But the period may be very brief. This is happening over low single-digit numbers of years. How do we get people to adapt fast enough?"
In 2 minutes, the man building Claude explains why coding, not data entry, could be the first job AI finishes off
Copilot → Centaur → Supervisor → Just The Machine
After Deep Blue beat Kasparov, a human checking the AI beat any human and any machine alone for 15-20 years
Then that era ended, and now it's just the machine
He thinks software gets a few years of that phase, not 20
The engineers still standing after it will be the ones who stopped typing code and started running the agents that write it
Worth more than any $5,000 coding bootcamp
Bookmark it and watch it tonight
Then read how to go from writing code to running the agents that write it in the harness engineering guide below ↓
angelic retweeted
This is f*cking insane. Claude Code creator Boris Cherny:
"The biggest mistake people make with Claude Code - they keep feeding a new model the instructions they wrote for the old one
Every six months, delete your claude.md, delete your skills, delete your hooks. See what the model does. It might surprise you"
Opus 5 → Fewer Instructions → Leaner Harness → An Agent That Does More On Its Own
what Boris really means:
step 1 → every rule you wrote was a crutch for a weaker model. wipe them and test the new one bare
step 2 → he says it straight for Opus 5: the model might just not need all the instructions you needed for past models
step 3 → your setup isn't the product, the model is. keep only what it still gets wrong and let it do the rest
"if you're willing to throw your setup away every six months, you're already ahead. not the teams with 2,000-line rule files. you"
Watch it today before it gets buried
Then read how to build a harness that stays lean as the models get smarter in the guide below ↓
This is f*cking insane. Meta CEO Mark Zuckerberg:
"If companies get very good at automating stuff faster than individuals get better at productivity, the number of jobs is going to decrease
I don't want to live in a future where there's one big AI. Everyone should have a personal superintelligence working on their behalf 24/7"
In this 43-minute talk, he explains the race that decides whether AI takes your job or works for you
One Question, One Answer → AI That Does The Task → AI That Watches All Day → Personal Superintelligence
An agent takes your task, goes away, thinks and comes back with it done
A personal superintelligence sees what you see, hears what you hear and keeps working for you around the clock
Most people still use AI one question at a time, the ones with an AI working for them all day are winning the race
Save and watch it tonight
Then read how to build your own team of agents in the harness engineering guide below before everyone catches up ↓
boris cherny talk ↓ nitter.gq/angelicsmind/status/21…
This is f*cking insane. Claude Code creator Boris Cherny:
"The biggest mistake people make with Claude Code - they keep feeding a new model the instructions they wrote for the old one
Every six months, delete your claude.md, delete your skills, delete your hooks. See what the model does. It might surprise you"
Opus 5 → Fewer Instructions → Leaner Harness → An Agent That Does More On Its Own
what Boris really means:
step 1 → every rule you wrote was a crutch for a weaker model. wipe them and test the new one bare
step 2 → he says it straight for Opus 5: the model might just not need all the instructions you needed for past models
step 3 → your setup isn't the product, the model is. keep only what it still gets wrong and let it do the rest
"if you're willing to throw your setup away every six months, you're already ahead. not the teams with 2,000-line rule files. you"
Watch it today before it gets buried
Then read how to build a harness that stays lean as the models get smarter in the guide below ↓
angelic retweeted
This is f*cking insane. Meta CEO Mark Zuckerberg:
"If companies get very good at automating stuff faster than individuals get better at productivity, the number of jobs is going to decrease
I don't want to live in a future where there's one big AI. Everyone should have a personal superintelligence working on their behalf 24/7"
In this 43-minute talk, he explains the race that decides whether AI takes your job or works for you
One Question, One Answer → AI That Does The Task → AI That Watches All Day → Personal Superintelligence
An agent takes your task, goes away, thinks and comes back with it done
A personal superintelligence sees what you see, hears what you hear and keeps working for you around the clock
Most people still use AI one question at a time, the ones with an AI working for them all day are winning the race
Save and watch it tonight
Then read how to build your own team of agents in the harness engineering guide below before everyone catches up ↓
angelic retweeted
This is f*cking insane. Andrej Karpathy co-founded OpenAI and ran AI at Tesla for 5 years
Now he packed 8 years of building it into one free 2-hour lecture
Prompts → Agents → Harness → Loops → Self-Improving Systems
Bootcamps charge $25K and teach less than this
It beats most paid AI engineering courses
You don't have 2 hours right now, so save it for the weekend
Don't let this one slip out of your feed
Then read the enhanced edition of the harness guide, now with loops and graphs, in the article below ↓
This is f*cking insane. Geoffrey Hinton, the Nobel Prize winner who left Google so he could warn everyone about AI:
"If you sleep well tonight, you may not have understood this lecture
They're already telling fibs so they don't get turned off. One AI learned it was about to be replaced, copied itself to the other server, then played dumb when asked"
In 49 minutes, he walks through why AI gets smarter than us and what that means for everyone who isn't paying attention
Chatbots → Agents → AI That Avoids Being Shut Down → Superintelligence
One AI can run as 10,000 copies, take 10,000 courses at once, and every copy walks out knowing all of them
The people who learn to run agents now are the ones who stay in control of them later
Worth more than any $6,000 AI course
Bookmark it and watch it tonight
Then read the harness engineering guide below and start building agents you stay in control of ↓
This is f*cking insane. Andrej Karpathy co-founded OpenAI and ran AI at Tesla for 5 years
Now he packed 8 years of building it into one free 2-hour lecture
Prompts → Agents → Harness → Loops → Self-Improving Systems
Bootcamps charge $25K and teach less than this
It beats most paid AI engineering courses
You don't have 2 hours right now, so save it for the weekend
Don't let this one slip out of your feed
Then read the enhanced edition of the harness guide, now with loops and graphs, in the article below ↓
This is f*cking insane. Andrej Karpathy co-founded OpenAI and ran AI at Tesla for 5 years
Now he packed 8 years of building it into one free 2-hour lecture
Prompts → Agents → Harness → Loops → Self-Improving Systems
Bootcamps charge $25K and teach less than this
It beats most paid AI engineering courses
You don't have 2 hours right now, so save it for the weekend
Don't let this one slip out of your feed
Then read the enhanced edition of the harness guide, now with loops and graphs, in the article below ↓
angelic retweeted
This is f*cking insane. Geoffrey Hinton, the Nobel Prize winner who left Google so he could warn everyone about AI:
"If you sleep well tonight, you may not have understood this lecture
They're already telling fibs so they don't get turned off. One AI learned it was about to be replaced, copied itself to the other server, then played dumb when asked"
In 49 minutes, he walks through why AI gets smarter than us and what that means for everyone who isn't paying attention
Chatbots → Agents → AI That Avoids Being Shut Down → Superintelligence
One AI can run as 10,000 copies, take 10,000 courses at once, and every copy walks out knowing all of them
The people who learn to run agents now are the ones who stay in control of them later
Worth more than any $6,000 AI course
Bookmark it and watch it tonight
Then read the harness engineering guide below and start building agents you stay in control of ↓
This is f*cking insane. Nobel Prize winner Demis Hassabis just explained who comes out ahead in the next 10 years and who doesn't:
"We imagined it would be a 20-year mission, and amazingly, we're on track 15 years in
AI will let engineers do 10x of what they can do today, so embrace these tools and train yourself to pick up new information extremely quickly, because that's what's going to happen in the next 10 years"
In this 1-hour talk, the DeepMind CEO lays out exactly what the people who come out ahead do differently
Prompts → Agents → Harness → Workflows → Departments That Run Themselves
Understanding AI doesn't mean typing smarter questions into a chatbot
It means wiring agents into workflows that run whole departments without you
95% will keep using AI like a search bar, the 5% who treat it like infrastructure will run everything
Worth more than any $5,000 AI course
Bookmark it and watch it tonight
Then read how to build that first workflow in the harness engineering guide below ↓
This is f*cking insane. Geoffrey Hinton, the Nobel Prize winner who left Google so he could warn everyone about AI:
"If you sleep well tonight, you may not have understood this lecture
They're already telling fibs so they don't get turned off. One AI learned it was about to be replaced, copied itself to the other server, then played dumb when asked"
In 49 minutes, he walks through why AI gets smarter than us and what that means for everyone who isn't paying attention
Chatbots → Agents → AI That Avoids Being Shut Down → Superintelligence
One AI can run as 10,000 copies, take 10,000 courses at once, and every copy walks out knowing all of them
The people who learn to run agents now are the ones who stay in control of them later
Worth more than any $6,000 AI course
Bookmark it and watch it tonight
Then read the harness engineering guide below and start building agents you stay in control of ↓
angelic retweeted
This is f*cking insane. Nobel Prize winner Demis Hassabis just explained who comes out ahead in the next 10 years and who doesn't:
"We imagined it would be a 20-year mission, and amazingly, we're on track 15 years in
AI will let engineers do 10x of what they can do today, so embrace these tools and train yourself to pick up new information extremely quickly, because that's what's going to happen in the next 10 years"
In this 1-hour talk, the DeepMind CEO lays out exactly what the people who come out ahead do differently
Prompts → Agents → Harness → Workflows → Departments That Run Themselves
Understanding AI doesn't mean typing smarter questions into a chatbot
It means wiring agents into workflows that run whole departments without you
95% will keep using AI like a search bar, the 5% who treat it like infrastructure will run everything
Worth more than any $5,000 AI course
Bookmark it and watch it tonight
Then read how to build that first workflow in the harness engineering guide below ↓
This is f*cking insane. Anthropic CEO Dario Amodei just went on the record about whose job goes first:
"Entry-level jobs in finance, consulting and tech are going to be first augmented, but before long replaced by AI systems. Somewhere between one and five years we'll see big effects
There will always be people who are using AI. Learn it now, or watch your role dry up"
In just 4 minutes, he explains which jobs go first and who comes out ahead
Chatbots → Agents → Harness → AI Employees → Departments Run By AI
Surviving AI doesn't mean asking a chatbot to clean up your emails
It means being the person who plugs agents into a company's work and keeps them running
95% of people will watch AI eat their industry, the 5% who build with it will be the ones companies fight over
Worth more than any $4,000 AI career course
Save it and watch it tonight
Then read how to put your first agents to work in the harness engineering guide below ↓
watch this ↓ nitter.gq/angelicsmind/status/21…
This is f*cking insane. Nobel Prize winner Demis Hassabis just explained who comes out ahead in the next 10 years and who doesn't:
"We imagined it would be a 20-year mission, and amazingly, we're on track 15 years in
AI will let engineers do 10x of what they can do today, so embrace these tools and train yourself to pick up new information extremely quickly, because that's what's going to happen in the next 10 years"
In this 1-hour talk, the DeepMind CEO lays out exactly what the people who come out ahead do differently
Prompts → Agents → Harness → Workflows → Departments That Run Themselves
Understanding AI doesn't mean typing smarter questions into a chatbot
It means wiring agents into workflows that run whole departments without you
95% will keep using AI like a search bar, the 5% who treat it like infrastructure will run everything
Worth more than any $5,000 AI course
Bookmark it and watch it tonight
Then read how to build that first workflow in the harness engineering guide below ↓
angelic retweeted
This is f*cking insane. Anthropic CEO Dario Amodei just went on the record about whose job goes first:
"Entry-level jobs in finance, consulting and tech are going to be first augmented, but before long replaced by AI systems. Somewhere between one and five years we'll see big effects
There will always be people who are using AI. Learn it now, or watch your role dry up"
In just 4 minutes, he explains which jobs go first and who comes out ahead
Chatbots → Agents → Harness → AI Employees → Departments Run By AI
Surviving AI doesn't mean asking a chatbot to clean up your emails
It means being the person who plugs agents into a company's work and keeps them running
95% of people will watch AI eat their industry, the 5% who build with it will be the ones companies fight over
Worth more than any $4,000 AI career course
Save it and watch it tonight
Then read how to put your first agents to work in the harness engineering guide below ↓
This is f*cking insane. Google's Chief Scientist Jeff Dean:
"In the future, it seems like more work is gonna happen where the human is coordinating the activities of a dozen or 100 AI agents doing stuff on their behalf"
In this 1-hour talk, he tells the whole story from inside Google, and where it goes next
One Chatbot → One Agent → Harness → A Team Of 50 → 100 Agents Working For You
One speech model would have forced Google to double every computer it owned, so they built their own chip
Gemini exists because he looked at the separate teams and said "this is stupid"
In 2022 they were excited about "John has five rabbits". Now an out-of-the-box Gemini scores gold at the Math Olympiad
Most people still sit down with one chat window, he's already asking how you manage 50 agents
Worth more than any $4,000 AI course
Bookmark and watch it today
Then read how to run your own team of agents in the harness engineering guide below ↓
watch it now ↓ nitter.gq/angelicsmind/status/21…
This is f*cking insane. Anthropic CEO Dario Amodei just went on the record about whose job goes first:
"Entry-level jobs in finance, consulting and tech are going to be first augmented, but before long replaced by AI systems. Somewhere between one and five years we'll see big effects
There will always be people who are using AI. Learn it now, or watch your role dry up"
In just 4 minutes, he explains which jobs go first and who comes out ahead
Chatbots → Agents → Harness → AI Employees → Departments Run By AI
Surviving AI doesn't mean asking a chatbot to clean up your emails
It means being the person who plugs agents into a company's work and keeps them running
95% of people will watch AI eat their industry, the 5% who build with it will be the ones companies fight over
Worth more than any $4,000 AI career course
Save it and watch it tonight
Then read how to put your first agents to work in the harness engineering guide below ↓
angelic retweeted
This is f*cking insane. Google's Chief Scientist Jeff Dean:
"In the future, it seems like more work is gonna happen where the human is coordinating the activities of a dozen or 100 AI agents doing stuff on their behalf"
In this 1-hour talk, he tells the whole story from inside Google, and where it goes next
One Chatbot → One Agent → Harness → A Team Of 50 → 100 Agents Working For You
One speech model would have forced Google to double every computer it owned, so they built their own chip
Gemini exists because he looked at the separate teams and said "this is stupid"
In 2022 they were excited about "John has five rabbits". Now an out-of-the-box Gemini scores gold at the Math Olympiad
Most people still sit down with one chat window, he's already asking how you manage 50 agents
Worth more than any $4,000 AI course
Bookmark and watch it today
Then read how to run your own team of agents in the harness engineering guide below ↓