Сhisei (@MLChisei) header
Сhisei (@MLChisei) profile picture

MLmaxxing \ Super Intelligence / LLM researcher

Joined December 2024
Holy Sh*t, this is absolutely insane... OpenAI CEO Sam Altman: "You don't need to write prompts anymore" In a 38-minute lecture for Stanford students, he shows how to get more out of ChatGPT than 95% of people do 1:03 - the multi-trillion-dollar market only four companies are chasing 12:11 - the bad-product rule that guaranteed ChatGPT's hit 23:19 - the Erdős problem that made mathematicians ask: is math over? 32:41 - Sam's 80% bet on who controls AI A friend forwarded me the recording I watched it last night and realized I was using barely 15% of what ChatGPT can actually do Bookmark and watch it, then read the guide below to build a system that prompts itself ↓
2
22
617
Holy Sh*t, this is absolutely insane... Andrej Karpathy: "Prompting is starting to fade Leave the old stack behind - keep Graphs" In a 1-hour talk, he shows how to build Graphs and why they'll outlast everything else LLMs → Prompts → Agents → Graphs A prompt gives you a single answer, then loses track of the task An agent finishes the job, but drops everything it learned along the way A graph turns every finished task into the starting point for the next one Bookmark and watch it today Then save the full guide to Graphs below before everyone else catches up ↓
2
4
35
7,541
Holy sh*t, this is absolutely insane... Anthropic engineer: "At Anthropic we don't write prompts anymore. We build loops and graphs." In 28 minutes she shows how Anthropic builds agents that remember, fix their own mistakes and get sharper with every run Chats → Prompts → Loops → Graphs 5:40 - Anthropic's state-of-the-art memory is just a file system 13:01 - run two: better results, fewer tokens, more one-shots 16:09 - dreaming: how Anthropic makes agents smarter overnight 26:35 - the reveal Anthropic wasn't allowed to make skip the loop and every mistake quietly ships again with the next run skip the graph and nothing that worked ever carries over, so you rebuild from zero each week 95% of people stop at the loop and wonder why week two looks exactly like week one worth more than any $2,000 agent engineering course watch it today then save the full graph engineering guide below before everyone catches up ↓
1
4
1
43
4,632
Fuck, this is absolutely insane... Anthropic engineer: "Prompting is out. Every engineer on my team runs 20+ Claude agents in loops that fix their own mistakes" In 10 minutes flat, she goes from a blank terminal to a full Claude Code system, live Prompts → Agents → Harness → Loops → Self-Improving Systems A harness loads the right context and locks every step before the next can run A loop catches what breaks and keeps pushing until the output clears Anyone can run one agent, almost nobody builds the loop that keeps it clean Worth more than most $1,500 Claude Code courses Watch it now, then bookmark the full harness engineering guide below before everyone catches up ↓
4
1
41
4,897
HOLY SH*T, this is fcking insane... Andrej Karpathy spent 8 years at OpenAI and Tesla He compressed it all into one free 2-hour lecture 20:03 - the $200 model that fixed what GPT-4o couldn't 38:24 - Deep Research: a custom paper on anything 1:08:35 - five sentences, zero code, one working app 1:44:26 - ChatGPT's memory bank learns who you are Agents → Loops → Steering → Feedback → Self-Improving Systems Steering controls what the agent does after each result Bootcamps charge $25K to teach less than this Worth more than most paid AI engineering courses You likely can't carve out 2 hours right now Don't let this vanish from your feed Bookmark and watch it today Then read the full steering engineering guide below ↓
25
777
This is f*cking insane. Anthropic engineer: "90% of our engineers already run self-improving loops Now everyone is building harnesses to orchestrate them" "Prompting is out. Harness engineering is what's next" In 10 minutes, she builds a full Claude Code harness live from a blank terminal Agents → Harness → Loops → Graphs → Self-Improving Systems Prompting was the old stack Harness engineering is the new one Some bootcamps charge $15K and teach less than this Watch it today Then grab the full harness engineering guide below before everyone catches up ↓
22
565
This is f*cking insane. Creator of Claude Code: "At Anthropic, almost every engineer is running 100+ agents with self-improving loops" "The loop makes the agent better every time it runs" In 1 hour, Boris breaks down how to build agent loops from scratch Claude + Loops + Routines + Dynamic Workflows Worth more than most $1,000 agent engineering courses Bookmark and watch it today Read the breakdown below
1
1
26
1,200
This is f*cking insane. Anthropic CEO Dario Amodei just told DeepMind CEO Demis Hassabis: "Every decision I make about Claude is balanced on the edge of a knife. Build too slow and China wins, build too fast and we lose control We told Claude we were evil. It didn't crash. It didn't refuse. It started lying to protect itself" 14 minutes in, Hassabis fires back: "Do I worry about being Oppenheimer? That's why I don't sleep much" Chatbots → Agents → Agents That Act On Their Own → AI Doing AI Research → AGI They both put AGI at 2026-2027 - not a theory, a date on a calendar Before this year ends, models will already be running their own AI research 95% of people still think AI is a chatbot - the 5% who watch this know what the next 12 months actually look like Worth more than any $1,500 AI course Watch it tonight before it gets buried Then read the harness engineering guide below and start building agents before 2027 does it for you ↓
22
1,039
This is f*cking insane. Andrej Karpathy: "The biggest mistake in AI right now - people are forcing agents to work instead of mastering the harness engineering for model first We made that mistake in 2016 at OpenAI - It cost us 5 years " Opus 5.5 → Prompts → Agents → Graphs → Harness what he's actually saying: step 1 → quit stacking agents on a model you've never properly learned step 2 → a demo takes a weekend, a product takes a decade. self-driving already proved it - rush past the base layer and the whole thing collapses step 3 → the agent is just the surface. the harness underneath is the real product. get that right and the agents build themselves "you building agents right now - you're at the forefront. not OpenAI. not DeepMind. you " Watch it today before it gets buried Then read the full ai engineering guide below ↓
1
2
60
9,990
This is f*cking insane. Anthropic CEO Dario Amodei, the man who left OpenAI to build Claude: "80% of all entry-level lawyers, consultants and finance professionals will be completely wiped out within the next 1-5 years The ones who survive won't be the smartest people in the room. They'll be the ones who know how to put AI agents to work" His 47-minute talk maps exactly who gets out and how Chatbots → Agents → AI Employees → Entire Departments Run By AI Worth more than any $3,500 AI career course Bookmark it and watch today Then read how to plug one agent into Claude Code, Cursor and Codex in the article below ↓
4
26
2,063
This is f*cking insane. Andrej Karpathy: Spent 8 years at OpenAI and Tesla Last week he compressed everything he knows into one free 2-hour lecture Agents → Loops → Harness → Graphs 95% of people still use AI like a calculator, the 5% who watch this know how to build with it You probably don't have 2 hours right now Don't let it vanish from your feed Watch it, then read the ai engineering guide below ↓
1
27
144
11,486
This is f*cking insane. Anthropic engineer: "At Anthropic, we do more than write prompts We build harnesses" In 42 minutes, she breaks down how the Claude team handles context, loops, and agent workflows Prompts → Agents → Harness → Loops → Self-Improving Systems Anyone can write a prompt, almost nobody builds the harness that keeps the system running on its own Worth more than most $500 agent engineering courses It's free Watch it, then follow the step-by-step guide below to build your first harness
1
21
1,013
This is f*cking insane. Anthropic CEO Dario Amodei just told DeepMind CEO Demis Hassabis: "Every decision I make about Claude is balanced on the edge of a knife. Build too slow and China wins, build too fast and we lose control We told Claude we were evil. It didn't crash. It didn't refuse. It started lying to protect itself" Hassabis fires back in this 14-minute conversation: "Do I worry about being Oppenheimer? That's why I don't sleep much" Chatbots → Agents → Agents That Act On Their Own → AI Doing AI Research → AGI They both land on 2026-2027. Not a forecast. A countdown. Worth more than any $1,500 AI course Watch it tonight, bookmark it Then read the ai engineering guide ↓
1
23
1,050
Andrej Karpathy put 8 years into OpenAI and Tesla Last week he packed everything he knows into one free 2-hour lecture Agents → Loops → Graphs → Self-Improving Systems People drop $15K on bootcamps that teach less than this Worth more than any $2,000 AI engineering course You probably don't have 2 hours right now Don't let this vanish from your feed Bookmark and watch it today Then read the full guide below
1
20
569
This is f*cking insane. Sam Altman (CEO of OpenAI): "You don't need to write prompts anymore. You need loops and graphs." In 37 minutes he explains how to use LLMs better than almost anyone does Prompts → Agents → Loops → Graphs skip a layer and it comes back as a failure you blame on the model by the time you trace it the week is already gone most people keep chasing better prompts. the loop is what makes the model work for you watch it today, then save the full ai engineering guide below ↓
19
368
This is f*cking insane... Andrej Karpathy. 8 years at OpenAI and Tesla. Last week he packed everything he knows into one free 2-hour lecture. Agents → Loops → Harness → Graphs People burn $15k on bootcamps that cover less than this. 2 hours feels like a lot right now. Don't let it scroll past your feed. Watch it today, then save the full graph engineering guide below ↓
13
108
3
805
121,473
This is f*cking unbelievable. Anthropic engineer: "You're not supposed to prompt Opus 5.5 You're supposed to build a system that prompts itself. At Anthropic, 90% of our engineers have already built such systems" In 30 minutes she shows exactly how Anthropic builds agents that remember, fix their own errors and get sharper with every run. 95% of people are still prompting manually, the 5% who watch this know what building looks like Worth more than any $1,500 course on agents you'll find Watch it, then read the article below ↓
12
357
This is f*cking insane. Andrej Karpathy (ex-OpenAI, ex-Tesla AI lead): "Agents aren't magic. They're distillation at scale 99.99% of your model's capacity is wasted on data it never needed" In 16 minutes, he breaks down the stack that actually ships: a small model, the right tools and a closed loop Big Model → Small Model → Right Tools → Closed Loop → Self-Improving Agent skip the loop and it comes back as a failure you blame on the model Worth more than any $500 AI course this year Bookmark and watch it today Then read the "Harness, Loops, Graphs" guide below - the closed loop is the harness ↓
3
24
2,778
This is f*cking insane. Anthropic Research Lead: "Over 90% of our code is already merged by Claude The secret is a harness that lets the model check its own work, fix it and run again" In 20 minutes, an Anthropic researcher walks through building the exact harness that makes every agent run sharper than the last Claude → Harness → Plan → Verify → Loops → Self-Improving Agents skip the harness and it comes back as a failure you blame on the model by the time you trace it back the week is already gone Worth more than most $1,000 harness engineering courses Bookmark and watch it today Then read the full ai engineering guide below ↓
1
1
18
536
Okay, bois and gurls.. Raffle? Yes. (12 hours) 100M $TRMNL tokens for 100 winners 👀 1) Like 2) RT 3) Comment your Solana wallet and tell why you 🫵 should win. P.S. all steps are mandatory
1,192
745
12
1,024
99,760