bottega.md (@bottegamd) header
bottega.md (@bottegamd) profile picture

obsessed with ai and technology u can do everything...

ocean of lights
Joined June 2023
Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)Photo by bottega.md (@bottegamd)
He ran Google for 10 years. Then he handed it back to the founders Now Eric Schmidt is giving away the easiest money play he sees, for free: "If you really want to make money, it's actually easy. Found an agentic AI company. Build an agent to do something If you build a better agent than anyone else, you can probably build one incredible company and own it yourself" In 30 seconds he shows why the next two years belong to whoever builds the best agent People pay $15K for bootcamps that teach less than this Everything you need to build that agent, free: Agent architecture langchain.com/blog?category_… Claude Code 101 anthropic.skilljar.com/claud… Claude Code in action anthropic.skilljar.com/claud… Prompt engineering platform.claude.com/docs/en/… Interactive prompt course github.com/anthropics/course… Claude.md + memory code.claude.com/docs/en/clau… Skills code.claude.com/docs/en/skil… MCP code.claude.com/docs/en/mcp Routines code.claude.com/docs/en/rout… Claude Code ultimate guide github.com/FlorianBruniaux/c… Awesome Claude Code github.com/hesreallyhim/awes… Anthropic academy anthropic.skilljar.com Official Claude Code docs code.claude.com/docs/en/over… All of this costs $0 Watch it before it gets buried ↓
2
28
583
bottega.md retweeted
"I haven't looked at code in a year" - Steve Yegge, ex-Google and ex-Amazon engineer In a live Q&A he lays out the ladder every engineer is climbing right now Most engineers still live in the IDE. One level up it's gone for good, and you talk to a "mayor" agent running a whole town of workers What it got him: 3,000 PRs in two months on Gastown, his open-source orchestrator Reading your own code now is like parking a Formula One car mid-race to look at the engine Watch this before you open your IDE tomorrow
1
1
20
707
"I haven't looked at code in a year" - Steve Yegge, ex-Google and ex-Amazon engineer In a live Q&A he lays out the ladder every engineer is climbing right now Most engineers still live in the IDE. One level up it's gone for good, and you talk to a "mayor" agent running a whole town of workers What it got him: 3,000 PRs in two months on Gastown, his open-source orchestrator Reading your own code now is like parking a Formula One car mid-race to look at the engine Watch this before you open your IDE tomorrow
1
1
20
707
bottega.md retweeted
This is f*cking insane. Andrej Karpathy, the man who co-founded OpenAI and ran AI at Tesla, just told the labs their giant models are memorizing garbage: "The internet is like 0.001% cognition and 99.99% of information is just garbage I think the cognitive core can be extremely small. I think even a billion, a billion suffices" In 17 minutes, he explains why the next big model is a small one that knows how to think and looks everything else up with tools Giant Model → Cognitive Core → Tools → A 1B Model That Thinks The labs are spending billions to make models remember the internet The model doesn't need to remember everything, it needs to know where to look and how to check This talk is worth more than most $1,000 AI courses Bookmark and watch it today Then read how to build the part around the model in the harness engineering guide below ↓
5
8
1
95
7,576
This is f*cking insane. Andrej Karpathy, the man who co-founded OpenAI and ran AI at Tesla, just told the labs their giant models are memorizing garbage: "The internet is like 0.001% cognition and 99.99% of information is just garbage I think the cognitive core can be extremely small. I think even a billion, a billion suffices" In 17 minutes, he explains why the next big model is a small one that knows how to think and looks everything else up with tools Giant Model → Cognitive Core → Tools → A 1B Model That Thinks The labs are spending billions to make models remember the internet The model doesn't need to remember everything, it needs to know where to look and how to check This talk is worth more than most $1,000 AI courses Bookmark and watch it today Then read how to build the part around the model in the harness engineering guide below ↓
5
8
1
95
7,576
bottega.md retweeted
This is f*cking insane. Google DeepMind CEO Demis Hassabis, the Nobel Prize winner behind AlphaFold, just called out every lab selling "PhD-level AI": "I think that's a nonsense. They have some capabilities that are PhD level, but they're not in general capable If you pose the question in a certain way, they can make simple mistakes with even high school maths" In about a minute, he puts AGI 5 to 10 years out and names what's still missing: continual learning and one or two breakthroughs Chatbots → PhD-Level Answers → High School Mistakes → Agents That Check Their Own Work The model can be brilliant and still miscount The people getting real work out of AI already assume it will slip, and build the checks around it The labs are arguing about who's smartest, almost nobody is building the part that catches the mistake before it reaches you Watch it today, then save the harness engineering guide below and build those checks yourself ↓
5
9
2
88
18,544
This is f*cking insane. Google DeepMind CEO Demis Hassabis, the Nobel Prize winner behind AlphaFold, just called out every lab selling "PhD-level AI": "I think that's a nonsense. They have some capabilities that are PhD level, but they're not in general capable If you pose the question in a certain way, they can make simple mistakes with even high school maths" In about a minute, he puts AGI 5 to 10 years out and names what's still missing: continual learning and one or two breakthroughs Chatbots → PhD-Level Answers → High School Mistakes → Agents That Check Their Own Work The model can be brilliant and still miscount The people getting real work out of AI already assume it will slip, and build the checks around it The labs are arguing about who's smartest, almost nobody is building the part that catches the mistake before it reaches you Watch it today, then save the harness engineering guide below and build those checks yourself ↓
5
9
2
88
18,544
bottega.md retweeted
This is f*cking insane. Google's chief scientist Jeff Dean, the engineer who built the systems Google still runs on: "We can compress chip design from 2 years to 3 months with reinforcement learning and new tooling It's essentially a specialized auto research loop for hardware" In just over 2 minutes, the man behind Google Brain and the TPU explains how a loop of agents does what used to take a team of engineers two years Engineers → Tools → Agents → An Auto Research Loop Design → Test → Score → Try Again, thousands of times a day This clip is worth more than most $1,000 AI courses Bookmark and watch it today Then read how to build your own loop in the harness engineering guide below ↓
1
1
36
3,041
This is f*cking insane. Turing Award winner Yann LeCun, one of the three godfathers of AI, just explained why he walked out of Meta: "The AI industry is completely LLM-pilled. Everybody is digging the same trench I cannot imagine that we can build agentic systems without those systems being able to predict the consequences of their actions" In under a minute, the man who invented the convolutional neural network says what every lab is getting wrong about agents Chatbots → LLMs → Agents → Agents That Know What Happens Before They Act An agent that only predicts the next word will happily delete the wrong folder An agent inside a harness checks what each step will do before it runs it Every lab is racing to build a bigger model, almost nobody is building the part that stops it from breaking things Save and watch it today Then read how to build that part in the harness engineering guide below ↓
12
22
1
123
21,314
This is f*cking insane. Google's chief scientist Jeff Dean, the engineer who built the systems Google still runs on: "We can compress chip design from 2 years to 3 months with reinforcement learning and new tooling It's essentially a specialized auto research loop for hardware" In just over 2 minutes, the man behind Google Brain and the TPU explains how a loop of agents does what used to take a team of engineers two years Engineers → Tools → Agents → An Auto Research Loop Design → Test → Score → Try Again, thousands of times a day This clip is worth more than most $1,000 AI courses Bookmark and watch it today Then read how to build your own loop in the harness engineering guide below ↓
70
This is f*cking insane. Google's chief scientist Jeff Dean, the engineer who built the systems Google still runs on: "We can compress chip design from 2 years to 3 months with reinforcement learning and new tooling It's essentially a specialized auto research loop for hardware" In just over 2 minutes, the man behind Google Brain and the TPU explains how a loop of agents does what used to take a team of engineers two years Engineers → Tools → Agents → An Auto Research Loop Design → Test → Score → Try Again, thousands of times a day This clip is worth more than most $1,000 AI courses Bookmark and watch it today Then read how to build your own loop in the harness engineering guide below ↓
1
1
36
3,041
bottega.md retweeted
This is f*cking insane. Turing Award winner Yann LeCun, one of the three godfathers of AI, just explained why he walked out of Meta: "The AI industry is completely LLM-pilled. Everybody is digging the same trench I cannot imagine that we can build agentic systems without those systems being able to predict the consequences of their actions" In under a minute, the man who invented the convolutional neural network says what every lab is getting wrong about agents Chatbots → LLMs → Agents → Agents That Know What Happens Before They Act An agent that only predicts the next word will happily delete the wrong folder An agent inside a harness checks what each step will do before it runs it Every lab is racing to build a bigger model, almost nobody is building the part that stops it from breaking things Save and watch it today Then read how to build that part in the harness engineering guide below ↓
12
22
1
123
21,314
bottega.md retweeted
This is f*cking insane. Ilya Sutskever co-founded OpenAI and built the models behind ChatGPT Now he's describing the one future he's actually afraid of: "Maybe every person will have an AI that does their bidding. It earns money for them, advocates for their needs, then writes a little report: here's what I've done The person says great, keep it up. But the person is no longer a participant" His only way out: people becoming part AI, through something like Neuralink++. And even he doesn't like that answer In this 96-minute interview, he explains why almost nobody will feel it coming until the report is already on their desk AI → Agents → AIs That Earn For You → Humans Who Just Sign Off The only way to stay a participant: be the one who runs the agents, not the one who reads their reports Worth more than any $6,000 AI course Save it and watch it before the weekend Then read how to build agents that still answer to you in the harness engineering guide below ↓
2
6
1
37
3,402
This is f*cking insane. 7 people walked out of OpenAI and built the $350B company behind Claude Dario Amodei, Daniela Amodei, Jared Kaplan, Jack Clark, Chris Olah, Tom Brown and Sam McCandlish sat at one table and told the whole story in one free 51-minute conversation They saw it coming in GPT-3: "If we don't do something soon, you're going to hit the point of no return" None of them wanted to start a company. Dario: "We felt like it was our duty" Jack Clark quit journalism for AI and got emails saying he was making "the worst mistake of your life" Jared Kaplan was a physics professor who "got bored" They all signed a pledge to give away 80% of what they make And Dario says Chris Olah will win a Nobel Prize in medicine. "I'm serious" Scaling Laws → GPT-3 → Anthropic → Claude → The Lab Everyone Is Racing Millions use Claude every day and have never heard the people who built it explain why Worth more than any $2,000 AI course You probably don't have 51 minutes right now. Save it before it disappears from your feed Then read how to put Claude to work in the harness engineering guide below ↓
2
2
1
32
968
This is f*cking insane. Ilya Sutskever co-founded OpenAI and built the models behind ChatGPT Now he's describing the one future he's actually afraid of: "Maybe every person will have an AI that does their bidding. It earns money for them, advocates for their needs, then writes a little report: here's what I've done The person says great, keep it up. But the person is no longer a participant" His only way out: people becoming part AI, through something like Neuralink++. And even he doesn't like that answer In this 96-minute interview, he explains why almost nobody will feel it coming until the report is already on their desk AI → Agents → AIs That Earn For You → Humans Who Just Sign Off The only way to stay a participant: be the one who runs the agents, not the one who reads their reports Worth more than any $6,000 AI course Save it and watch it before the weekend Then read how to build agents that still answer to you in the harness engineering guide below ↓
39
This is f*cking insane. Ilya Sutskever co-founded OpenAI and built the models behind ChatGPT Now he's describing the one future he's actually afraid of: "Maybe every person will have an AI that does their bidding. It earns money for them, advocates for their needs, then writes a little report: here's what I've done The person says great, keep it up. But the person is no longer a participant" His only way out: people becoming part AI, through something like Neuralink++. And even he doesn't like that answer In this 96-minute interview, he explains why almost nobody will feel it coming until the report is already on their desk AI → Agents → AIs That Earn For You → Humans Who Just Sign Off The only way to stay a participant: be the one who runs the agents, not the one who reads their reports Worth more than any $6,000 AI course Save it and watch it before the weekend Then read how to build agents that still answer to you in the harness engineering guide below ↓
2
6
1
37
3,402
bottega.md retweeted
This is f*cking insane. 7 people walked out of OpenAI and built the $350B company behind Claude Dario Amodei, Daniela Amodei, Jared Kaplan, Jack Clark, Chris Olah, Tom Brown and Sam McCandlish sat at one table and told the whole story in one free 51-minute conversation They saw it coming in GPT-3: "If we don't do something soon, you're going to hit the point of no return" None of them wanted to start a company. Dario: "We felt like it was our duty" Jack Clark quit journalism for AI and got emails saying he was making "the worst mistake of your life" Jared Kaplan was a physics professor who "got bored" They all signed a pledge to give away 80% of what they make And Dario says Chris Olah will win a Nobel Prize in medicine. "I'm serious" Scaling Laws → GPT-3 → Anthropic → Claude → The Lab Everyone Is Racing Millions use Claude every day and have never heard the people who built it explain why Worth more than any $2,000 AI course You probably don't have 51 minutes right now. Save it before it disappears from your feed Then read how to put Claude to work in the harness engineering guide below ↓
2
2
1
32
968
This is f*cking insane. The godmother of AI, Fei-Fei Li, the Stanford professor who built ImageNet: "Language is fundamentally a purely generated signal. You don't go out in nature and there's words written in the sky for you There is a 3D world out there that follows the laws of physics. To back that information out and generate it is fundamentally a different problem" In 1 minute, she explains why the AI you type into can't understand the real world from words alone Words → LLMs → World Models → AI That Understands Physics An LLM learns from text, and text is something humans made up to describe the world A world model learns from the world itself: space, materials, gravity, the things no sentence fully captures Everyone is still arguing over which chatbot writes better, almost nobody is watching the race that comes after it Worth more than any $10,000 AI course Bookmark it and watch it tonight Then read how to build agents for what comes after the chatbot in the harness engineering guide below ↓
Your agent passed every test The issue is still open A better prompt won't fix it, because the model isn't the broken part I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
2
5
74
4,939
This is f*cking insane. 7 people walked out of OpenAI and built the $350B company behind Claude Dario Amodei, Daniela Amodei, Jared Kaplan, Jack Clark, Chris Olah, Tom Brown and Sam McCandlish sat at one table and told the whole story in one free 51-minute conversation They saw it coming in GPT-3: "If we don't do something soon, you're going to hit the point of no return" None of them wanted to start a company. Dario: "We felt like it was our duty" Jack Clark quit journalism for AI and got emails saying he was making "the worst mistake of your life" Jared Kaplan was a physics professor who "got bored" They all signed a pledge to give away 80% of what they make And Dario says Chris Olah will win a Nobel Prize in medicine. "I'm serious" Scaling Laws → GPT-3 → Anthropic → Claude → The Lab Everyone Is Racing Millions use Claude every day and have never heard the people who built it explain why Worth more than any $2,000 AI course You probably don't have 51 minutes right now. Save it before it disappears from your feed Then read how to put Claude to work in the harness engineering guide below ↓
42
bottega.md retweeted
This is f*cking insane. The godmother of AI, Fei-Fei Li, the Stanford professor who built ImageNet: "Language is fundamentally a purely generated signal. You don't go out in nature and there's words written in the sky for you There is a 3D world out there that follows the laws of physics. To back that information out and generate it is fundamentally a different problem" In 1 minute, she explains why the AI you type into can't understand the real world from words alone Words → LLMs → World Models → AI That Understands Physics An LLM learns from text, and text is something humans made up to describe the world A world model learns from the world itself: space, materials, gravity, the things no sentence fully captures Everyone is still arguing over which chatbot writes better, almost nobody is watching the race that comes after it Worth more than any $10,000 AI course Bookmark it and watch it tonight Then read how to build agents for what comes after the chatbot in the harness engineering guide below ↓
Your agent passed every test The issue is still open A better prompt won't fix it, because the model isn't the broken part I broke down the harness, loop and graph around it: what each one owns, how they nest, and which one to fix first ↓
2
5
74
4,939