0xvexran

building in ai | ai content

dm open

[0xvexran @0xvexran]

@shiqway92 book

[0xvexran @0xvexran]

@kocer_eth grok bot imba

[0xvexran @0xvexran]

@0x_kaize shock bro

[0xvexran @0xvexran]

@0xboan genius setup

[0xvexran @0xvexran]

@FLOOF_Crypto BlackRock genius

[0xvexran @0xvexran]

Anthropic paid this engineer $250K–$750K a year for one skill: building harnesses for multi-agent systems

In a 15-minute workshop, he walks through building one from the ground up

AI → Agents → Harness → Loops → Graphs

step 1 → start with the Claude Agent SDK - it handles loops, context, and sandboxing for you

step 2 → keep the brain separate from the hands - reasoning stays in one place, tools run in a sandbox, first token comes 60% faster

step 3 → run it server-side and log every single step - close your laptop, it keeps going, and crashes just resume from the log

step 4 → make failure cost nothing - retry dead sandboxes and replay lost context instead of starting from scratch

step 5 → turn yesterday's logs into fresh memory and skills - the harness comes back smarter than it left

Anthropic actually calls this "dreaming"

Most people spend weeks building this from scratch

You won't have to

Bookmark it, watch it

[rari @0xwhrrari]

https://t.co/3ANIWs48lJ

[0xvexran @0xvexran]

@FLOOF_Crypto @jpmorgan a very smart man

[0xvexran @0xvexran]

@shiqway92 thanks bro

[0xvexran @0xvexran]

Sam Altman, CEO of OpenAI:

"You don't need to write your own prompts anymore"

In under 35 minutes, he breaks down how to get more out of ChatGPT than most users ever do

He originally gave this talk to a group of Stanford students

A friend passed me the recording

Watched it last night and realized I'd been using less than 15% of what ChatGPT is actually capable of

Watch it, then check the guide below for building a system that prompts itself

[rari @0xwhrrari]

https://t.co/3ANIWs48lJ

[0xvexran @0xvexran]

@shiqway92 cockroach solo

[0xvexran @0xvexran]

Andrew Ng - the guy who founded Google Brain - just said something bold:

"Prompting has six months left. Graphs are taking over."

He proved it in a 2-hour Stanford lecture, walking through exactly how self-operating, self-improving agents get built.

The path: Agents → Loops → Graphs → Self-Improving Systems

Ten minutes in, you'll already know more than most $500 courses ever teach you.

Free > most paid agent engineering programs, honestly.

Go watch it.

Then hit the guide below for the full breakdown on systems that improve on their own.

[rari @0xwhrrari]

https://t.co/3ANIWs48lJ

[0xvexran @0xvexran]

@FLOOF_Crypto amazing article

[0xvexran @0xvexran]

Andrej Karpathy put in 8 years across OpenAI and Tesla

Last week, he packed everything he's learned into one free 2-hour lecture

People pay $15k for bootcamps that cover half of what's in here

You probably don't have 2 hours to spare right now

Don't let this get buried in your feed

Watch it, then read the guide below and build your first loop

[rari @0xwhrrari]

https://t.co/3ANIWs48lJ

[0xvexran @0xvexran]

A buddy of mine pulls in something like $1.1M a year working as an engineer at Anthropic

I asked him how he got so sharp at prompting

He sent over a video that was never meant to go public

The exact prompting playbook their internal team runs on

There's nothing out there on prompting that beats this

Watched the whole thing last night

Halfway in, it hit me - I'd been using Claude wrong this entire past year and a half

Watch it. Save it.

[rari @0xwhrrari]

https://t.co/3ANIWs48lJ

[0xvexran @0xvexran]

@0xtoriin I think so too

[0xvexran @0xvexran]

Anthropic engineer: "At Anthropic, we don't write prompts anymore. We build loops"

In 27 minutes, she shows how the Claude team builds loops that prompt themselves

If this cost $250, people would call it the best agents course of the year

It's free

Watch it, then use the step-by-step guide below to build your first loop

[rari @0xwhrrari]

https://t.co/3ANIWs48lJ

[0xvexran @0xvexran]

@0xtoriin Grok Bot monster