sendme (@sendme_ai) header
sendme (@sendme_ai) profile picture
Joined July 2026
Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)Photo by sendme (@sendme_ai)
Pinned Tweet
This is f*cking insane. Anthropic is now a $1.5 trillion public company. Months ago, Dario Amodei explained the model they refused to release: "Mythos was much better than human software engineers at finding ways that code could be exploited One company found and fixed more bugs with it in a single week than in the entire previous year" In 1 hour, he walks through the actual frontier of AI, and why some models are too good to ship Boilerplate Code → Better Than Engineers → Too Dangerous to Release → $1.5 Trillion Everyone is still debating whether AI can write code. That debate is over Bookmark and watch it today Then read how to put Claude to work: 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
38
1,220
This is f*cking insane. Matt Palmer, the guy behind the most watched Grok Bot videos on X, recorded a full Claude Code masterclass: "Run Claude Code in the cloud, start parallel agents from your phone, and ship while you're away from your desk" In 15 minutes, he covers dev vs prod, building locally vs in the cloud, and running parallel agents straight from your phone Local → Cloud → Parallel Agents → Shipping From Your Phone Worth more than any $1,500 Claude Code course Bookmark and watch it today Then read how to set up your own loops: 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
37
895
This is f*cking insane. GTA 6 just dropped, and people are already calling it the last game ever built without AI. Demis Hassabis started his career designing video games. Now he runs Google DeepMind: "I would have left AI in the lab longer" In 65 minutes, the Nobel Prize winner explains the hardest problem AI has already solved, what humans can do that AI won't, and what he wants his legacy to be Video Games → DeepMind → AlphaFold → Nobel Prize → AGI Worth more than any $2,000 AI course Bookmark and watch it tonight Then read how to put today's AI to work: 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
3
65
7,693
This is f*cking insane. Google Brain founder Andrew Ng: "Prompting will be dead in 6 months. Graphs are what's replacing it" In 2 hours at Stanford, he shows how to build agents that work and improve on their own Prompts → Agents → Graphs → Agents That Improve Themselves The first 10 minutes cover what most $500 courses never do Bookmark and watch it tonight Then read how to build a system that improves itself: 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 ↓
5
59
4,836
This is f*cking insane. Stanford put out a free lecture on how ChatGPT and Claude actually work In under 2 hours, it goes through everything under the hood, from how a model reads your text to how it learns to reason Tokens → Training → Fine-Tuning → Reasoning → Agents Useful whether you've never touched AI or use it every single day Skip one night of Netflix for this Worth more than any $2,000 AI course Bookmark and watch it tonight Then read how to get 100% out of Claude: 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
1
46
1,285
This is f*cking insane. Boris Cherny, the man who built Claude Code: "I don't prompt Claude anymore. I have loops that figure out what to do My job is to create the loops" In 30 minutes, he breaks down his daily Claude Code setup, step by step Prompt → Chat Box → Loops → Loops That Run on Their Own The person who built the tool doesn't use it the way most people think Worth more than most $1,500 Claude courses Bookmark and watch it today Then read how those loops are 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
4
52
1,505
This is f*cking insane. The engineer who built Claude Code just showed how he actually prompts it: "Most people write a better sentence and hope. I write the files the model reads before it starts" In 28 minutes, he covers CLAUDE.md files, memory shortcuts, parallel sessions and the prompting patterns his team uses every day Prompt → CLAUDE.md → Memory → Parallel Sessions → Claude That Knows Your Project I've seen $300 courses that don't cover what he shows in the first 10 minutes Bookmark and watch it today Then read what comes after the prompt: 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
8
1
227
22,140
This is f*cking insane. Anthropic engineers just gave away a 90-minute course on building agents from scratch: "Keep the brain and the hands separate. That one decision is why our agents don't fall apart" In 94 minutes, they build a research agent step by step, then agents that run on a schedule and "dream" between tasks Brain → Hands → Research Agent → Scheduled Agents → Evals and CI This is the architecture behind Claude Managed Agents Worth more than every $500 agent tutorial on the internet Bookmark and watch it this week Then read the short version: 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
37
797
sendme retweeted
This is f*cking insane... Stop paying Claude Opus $6,880 to read your own documents XERJ just opened the waitlist for a model that doesn't generate a single word You give it a file and a question, it highlights the exact line with the answer, or says "no answer" The same 1M questions: → GPT-6 Astra - $17,200 → Claude Opus 5.5 - $6,880 → Claude Haiku 5.5 - $172 → XERJ Reader - $6.50 $0.01 per 1M input tokens, output is free The answer lands in 80 ms, Opus needs about 5 seconds just to start typing It can't make up a number, it can only return text that is already in your file I put the whole model on one page, bookmark it before your next API bill Get on the waitlist while it's open ↓
🤝 Paid partnership
I've built a new AI model that is not allowed to generate text. Instead, it produces a verifiable program that answers your prompt using the original reference. XERJ Reader-1 is a neurosymbolic model for working with data. It points to the exact words in your document, writes a small program over them, and exact code computes the answer: • 20-90x faster • 10-1,000x cheaper (output tokens free) • every answer is a program you can replay The shortest path to automating data work.
9
2
67
3,141
This is f*cking insane. Andrej Karpathy spent 8 years building AI at OpenAI and Tesla Then he packed everything he knows into one free 2-hour lecture Agents → Harness → Loops → Graphs People pay $15K for bootcamps that teach less than this You probably don't have 2 hours right now Don't let it disappear from your feed Bookmark it and watch it this week Then read the short version: 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
39
858
This is f*cking insane. Stanford just put out a free 2.5-hour course on building an LLM from scratch: "Most people use ChatGPT and Claude every day and have no idea what happens after they hit enter" In 2.5 hours, it goes from the first token to a full model you could train yourself Tokenization → Decoding → Training Pipeline → Architecture → Your Own LLM AI labs pay $750K a year for exactly this knowledge Worth more than a 2-year master's degree Bookmark it and find 2 hours today Then read the loop that turns a model like this into agents that 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 ↓
4
12
128
7,429
This is f*cking insane. Stanford professor Andrew Ng just gave away a 95-minute course on harness engineering: "Prompting will die in 7 months Harnesses are what replace it" In 95 minutes, he builds a harness that lets agents plan, execute, verify and improve on their own Prompts → Agents → Harness → Loops → Graphs → Self-Improving Systems Most people are still tuning prompts while the real work already moved into the harness Worth more than a $15K bootcamp Bookmark and watch it today Then read the short version: Harness, Loops, Graphs ↓
1
24
This is f*cking insane. Stanford professor Andrew Ng just gave away a 95-minute course on harness engineering: "Prompting will die in 7 months Harnesses are what replace it" In 95 minutes, he builds a harness that lets agents plan, execute, verify and improve on their own Prompts → Agents → Harness → Loops → Graphs → Self-Improving Systems Most people are still tuning prompts while the real work already moved into the harness Worth more than a $15K bootcamp Bookmark and watch it today Then read the short version: 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
1
47
1,300
This is f*cking insane. Google just put out a free 2-hour course on harness, loop and graph engineering: "Most people trust an agent because its answer looks right. We make it prove it" From one prompt to 100+ agents that test, judge and fix their own work Harness → First Agent → Evals → Loops → A Graph That Improves While You Sleep Fable 5.2 is out, and the model was never the problem. A setup with no checks is Worth more than any $2,500 agentic engineering course Save it and watch it tonight Then read the short version of the same system: 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
1
45
734
sendme retweeted
This is f*cking insane. Stanford just put out a free 2.5-hour course on building an LLM from scratch: "Most people use ChatGPT and Claude every day and have no idea what happens after they hit enter" In 2.5 hours, it goes from the first token to a full model you could train yourself Tokenization → Decoding → Training Pipeline → Architecture → Your Own LLM AI labs pay $750K a year for exactly this knowledge Worth more than a 2-year master's degree Bookmark it and find 2 hours today Then read the loop that turns a model like this into agents that 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 ↓
4
12
128
7,429
This is f*cking insane. Anthropic engineer: "Prompting is basically over. Almost everyone on our team runs self-improving loops now, and graphs are next" In 10 minutes, she opens an empty terminal and builds her full Claude Code setup live Empty Terminal → Agents → Loops → Graphs → A System That Upgrades Itself The people who build Opus 5.5 don't write prompts. They build the system the prompts come from Worth more than most $2,000 agent courses Bookmark and watch it today Then read the loop that stops you from checking every step yourself: 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
38
650
This is f*cking insane. Ilya Sutskever, OpenAI co-founder and the mind behind ChatGPT, on the future he's afraid of: "Every person will have an AI that earns money for them and writes a little report: here's what I did today The person says great, keep going. But the person is no longer a participant" In this 97-minute interview, he explains why almost nobody feels what's coming until it's already here Chat → Agents → AIs That Earn For You → People Who Only Read The Reports Stay a participant: run the agents yourself instead of reading their reports Worth more than any $4,000 AI course Bookmark it and watch it tonight Then read how an AI employee runs QC for motion engineering with Opus 5.5 ↓
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 ↓
2
35
635
sendme retweeted
This is f*cking insane. SpaceXAI just put out its best 1-hour Grok Bot course for free: "Most people build one bot and stop. The ones who win give every bot a role and let them run 24/7" In 62 minutes, it goes from your very first bot to a full system that runs without you First Bot → A Role for Every Bot → 24/7 Bots → Parallel Agents → Fully Autonomous System Every step is something you can set up the same evening Worth more than a $1,500 automation bootcamp Bookmark and watch it today Then read "Your AI video is not bad. Your checks are.", the QC layer of motion engineering with Opus 5.5 ↓
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
2
35
600
This is f*cking insane. SpaceXAI just put out its best 1-hour Grok Bot course for free: "Most people build one bot and stop. The ones who win give every bot a role and let them run 24/7" In 62 minutes, it goes from your very first bot to a full system that runs without you First Bot → A Role for Every Bot → 24/7 Bots → Parallel Agents → Fully Autonomous System Every step is something you can set up the same evening Worth more than a $1,500 automation bootcamp Bookmark and watch it today Then read "Your AI video is not bad. Your checks are.", the QC layer of motion engineering with Opus 5.5 ↓
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
2
35
600
sendme retweeted
HOLY SH*T. Nobel Prize winner Geoffrey Hinton, the Godfather of AI: "Anybody who uses a chatbot regularly knows they understand We have to think that they're very like us. They're beings like us" If that's true, you'd better understand how they work. Start here, free: Karpathy's Neural Networks: Zero to Hero karpathy.ai/zero-to-hero.htm… Hugging Face LLM course huggingface.co/learn/llm-cou… Practical deep learning course.fast.ai/ Google's machine learning crash course developers.google.com/machin… Building effective agents anthropic.com/research/build… Grok API docs docs.x.ai/overview All of it costs $0 Bookmark it Then read the "6 Bots, 200+ Agents" Grok Bot setup below ↓
Your Grok Bot is not expensive Your setup is 75% of my Grok Bot tokens went to re-reading old chat. About 1% went to actual work I broke down the 6-bot setup the Grok Bot team runs, and the 9 leaks eating your bill ↓
2
1
1
39
676