realno

Web3 Research | AI Research | dm open

[realno @rea1no1]

This is f*cking insane. Anthropic engineer:

"99% of people still use Fable 5.2 like Google

The top 1% build a harness and let swarms of Claude agents work, coordinate and improve on their own. I run 50+ agents inside one"

In this 27-minute workshop, he shows how to get the most out of Claude Code for the least money

Model → Tools → Harness → Closed Loops → Self-Improving System

A Chief agent and a PM agent run the whole thing. He checks the result, not the work

Worth more than every $1,000 vibe-coding course combined

Bookmark and watch it today

Then read how the harness is built: Harness, Loops, Graphs ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

https://t.co/aQzTYaFVK9

[realno @rea1no1]

This is f*cking insane. Anthropic engineer:

"Stop prompting Fable 5.2. Build the system that prompts it for you

90% of our engineers already run one. It remembers, fixes its own mistakes and gets sharper every run"

In 31 minutes, she walks through exactly how Anthropic builds agents that improve without you

Prompt → Agent → Memory → Self-Correction → A System That Prompts Itself

The model got smarter this week. The people still typing prompts did not

Worth more than any $1,500 agent course

Bookmark and watch it today

Then read how that system is wired: Harness, Loops, Graphs ↓

[realno @rea1no1]

This is f*cking insane. The head of Claude Code at Anthropic:

"90% of our engineers run dozens or hundreds of agents on Fable 5.2

Nobody prompts them one by one. The way you do it is graph engineering"

In 35 minutes, he shows how one engineer at Anthropic now does the work of a whole team, and where it goes next

One Agent → Dozens Of Agents → Graphs → Hundreds Of Agents → Self-Improving System

Worth more than most $1,500 agent engineering courses

Bookmark and watch it today

Then read the graph layer in full: Harness, Loops, Graphs ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

This is f*cking insane. Anthropic engineer:

"Stop prompting Fable 5.2. Build the system that prompts it for you

90% of our engineers already run one. It remembers, fixes its own mistakes and gets sharper every run"

In 31 minutes, she walks through exactly how Anthropic builds agents that improve without you

Prompt → Agent → Memory → Self-Correction → A System That Prompts Itself

The model got smarter this week. The people still typing prompts did not

Worth more than any $1,500 agent course

Bookmark and watch it today

Then read how that system is wired: Harness, Loops, Graphs ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

This is f*cking insane. Cursor CEO Michael Truell:

"We were four 20-somethings running the whole service with 5 people, and we were already a huge share of the AI labs' API revenue

AI coding just had its iPod moment. The iPhone moment is next, and most engineers won't see it coming"

In 28 minutes, he shows how 4 MIT kids with zero marketing built the fastest growing software company ever

Failed CAD Startup → AI Editor From Scratch → $2B A Year → A $60B 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 tonight

Then read how to build the team of agents that writes your code for you: Harness, Loops, Graphs ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

this Anthropic paper is f*cking insane

it explains why your agent passes every test and the issue is still open.

9 rules for what "done" actually means, plus a verifier prompt you can paste as is.

paste it once and your agent stops grading its own homework.

the crazy part is the tests were never the proof. 99% of people ship whatever turns green.

the 7 levels from an agent you babysit to a fleet that checks itself are in the article below ↓

save it before it gets buried.

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

this Anthropic paper is f*cking insane

it's pretty much the QC system top AI teams wrap around Opus 5.5 so the same bug never costs them twice.

one log, 10 defect classes, 9 rules and a job description you can paste as is.

paste it into your agent once and it stops marking its own work "done".

the crazy part is the model was never the problem. 99% of people fix the same bug every project because nothing remembers the fix.

the real run, 100+ defects across 10 Opus 5.5 films, is in the article below ↓

save it before it gets buried.

[m4lka @m4lkxx]

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 ↓ https://t.co/i7RRo3bIg2

[realno @rea1no1]

this Anthropic paper is f*cking insane

it's pretty much the 7-level system Anthropic engineers use to turn Opus 5.5 from a chat into an agent that actually finishes the job.

every layer, every formula and a template you can paste as is.

paste it into Opus 5.5 once and it stops saying "done" before a single test has passed.

the crazy part is the fix was never a better prompt. 99% of people are still the loop themselves: it writes, they check, they type again.

the same system applied to motion engineering with Opus 5.5 is in the article below ↓

save it before it gets buried.

[realno @rea1no1]

this Anthropic paper is f*cking insane

it's pretty much the 7-level system Anthropic engineers use to turn Opus 5.5 from a chat into an agent that actually finishes the job.

every layer, every formula and a template you can paste as is.

paste it into Opus 5.5 once and it stops saying "done" before a single test has passed.

the crazy part is the fix was never a better prompt. 99% of people are still the loop themselves: it writes, they check, they type again.

the same system applied to motion engineering with Opus 5.5 is in the article below ↓

save it before it gets buried.

[m4lka @m4lkxx]

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 ↓ https://t.co/i7RRo3bIg2

[realno @rea1no1]

This is f*cking insane. Anthropic CEO Dario Amodei:

"In the next 2 to 5 years, AI will hit the job market faster and broader than anything we've ever seen"

He's not forecasting. He says it's already happening

In this 22-minute talk, he explains how to come out on the right side of it

Chatbots → Agents → AI Employees → A Job Market Rebuilt

You can't adapt to something you ignore

Worth more than any $1,000 AI career course

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[realno @rea1no1]

This is f*cking insane. OpenAI engineer:

"85% of our engineers are running hundreds of agents

The way you do it is graph engineering"

In this 25-minute lecture, he explains how one engineer now does the work of a whole team

One Engineer → Hundreds of Agents → Graphs → The Work of a Whole Team

Same idea at OpenAI and at Anthropic. That's not a coincidence

Worth more than most $1,000 agent courses

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

@0xg4tsby bookmarked 👍

[realno @rea1no1]

This is f*cking insane. Anthropic CEO Dario Amodei:

"In the next 2 to 5 years, AI will hit the job market faster and broader than anything we've ever seen"

He's not forecasting. He says it's already happening

In this 22-minute talk, he explains how to come out on the right side of it

Chatbots → Agents → AI Employees → A Job Market Rebuilt

You can't adapt to something you ignore

Worth more than any $1,000 AI career course

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

This is f*cking insane. Anthropic research lead:

"99% of our engineers run swarms of 300+ self-improving agents

Close the loop. Give the model a way to check its own work"

In a 20-minute session, she shows how to build agents that get better every time they run

Claude → Loops → Plan Mode → Verification → Self-Improving Agents

The model checking itself is the whole trick

Worth more than most $1,000 agent courses

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[realno @rea1no1]

This is f*cking insane. Anthropic research lead:

"99% of our engineers run swarms of 300+ self-improving agents

Close the loop. Give the model a way to check its own work"

In a 20-minute session, she shows how to build agents that get better every time they run

Claude → Loops → Plan Mode → Verification → Self-Improving Agents

The model checking itself is the whole trick

Worth more than most $1,000 agent courses

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

Anthropic engineers get paid $500K+ a year

This 30-minute video is how their core team actually prompts Claude

Halfway through, you realize you've been using it wrong for two years

Vague Ask → Clear Role → Examples → Self-Check → Output You Don't Have to Fix

Bookmark it and watch it tonight

Then read the "Harness, Loops, Graphs" guide below ↓

[realno @rea1no1]

Anthropic engineers get paid $500K+ a year

This 30-minute video is how their core team actually prompts Claude

Halfway through, you realize you've been using it wrong for two years

Vague Ask → Clear Role → Examples → Self-Check → Output You Don't Have to Fix

Bookmark it and watch it tonight

Then read the "Harness, Loops, Graphs" guide below ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

This is f*cking insane. Anthropic just put out Boris Cherny and Cat Wu, the people behind Claude Code, on how it escaped engineering

It started as a coding tool. Now the whole company runs on it, not just the engineers

11 minutes, straight from Anthropic's own account

Claude Code → Engineers → The Whole Company → Claude Tag

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[realno @rea1no1]

This is f*cking insane. Anthropic just put out Boris Cherny and Cat Wu, the people behind Claude Code, on how it escaped engineering

It started as a coding tool. Now the whole company runs on it, not just the engineers

11 minutes, straight from Anthropic's own account

Claude Code → Engineers → The Whole Company → Claude Tag

Bookmark and watch it today

Then read the "Harness, Loops, Graphs" guide below ↓

[m4lka @m4lkxx]

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 ↓ https://t.co/4RacdXA4nE

[realno @rea1no1]

@fomoebka bookmark