slayr

AI Engineer

[slayr @slayrsays]

No f*cking way, Anthropic just dropped a free 4-hour course on building AI agents that check their own work

From one prompt to agents that prove every change before you ever see it:

0% → 00:13 - The prompting playbook Anthropic engineers actually use
30% → 48:48 - Why your agent gets worse every time you add a feature
45% → 1:38:15 - Make Claude verify its own work, natively
75% → 1:53:20 - Make it record the proof of every check
100% → 2:46:19 - A verifier routine that checks your deploys while you sleep

Prompt → Agent → Verifier → Proof → Routines → Code That Ships Checked While You Sleep

The Claude Code engineering lead at 3:26:26: coding isn't the bottleneck anymore, verification is

Most people still check every line Claude writes by hand

Worth more than any $4,000 agentic engineering bootcamp

Bookmark and watch it today

[distort @distortgeekin]

SpaceXAI's Lauren Tan stopped reading her agents' work and her merged PRs went 13x

The trick wasn't a smarter model, it was a second agent whose only job is to say no

6 levels of done and a CHECKS.md you can copy ↓ https://t.co/enkux2Y8N0

[slayr @slayrsays]

Holy sh*t, Microsoft CEO Satya Nadella just admitted the software Microsoft sells will collapse in the agent era:

"Business apps are just databases with logic on top, and all of that logic is moving into the agents

Open Excel with Copilot and you're not using a spreadsheet anymore, you're managing a data analyst"

In 2 minutes, he explains how Microsoft goes from selling you apps to selling you agents that run every app at once

Apps → Copilot → Agents → One Agent Layer Over Every Database → Software Nobody Has To Open

He says Microsoft is already pulling customers off old CRM backends onto agents, and finance and ops are next

The app was never the product, the logic was, and now someone has to draw which agent owns which piece

Most people are still learning one more app

Worth more than any $3,500 AI agent course

Bookmark it, watch it today

[slayr @slayrsays]

Holy sh*t, Microsoft CEO Satya Nadella just admitted the software Microsoft sells will collapse in the agent era:

"Business apps are just databases with logic on top, and all of that logic is moving into the agents

Open Excel with Copilot and you're not using a spreadsheet anymore, you're managing a data analyst"

In 2 minutes, he explains how Microsoft goes from selling you apps to selling you agents that run every app at once

Apps → Copilot → Agents → One Agent Layer Over Every Database → Software Nobody Has To Open

He says Microsoft is already pulling customers off old CRM backends onto agents, and finance and ops are next

The app was never the product, the logic was, and now someone has to draw which agent owns which piece

Most people are still learning one more app

Worth more than any $3,500 AI agent course

Bookmark it, watch it today

[distort @distortgeekin]

Everyone builds the agents first, then wonders why they step on each other

Draw the work first: steps, handoffs, decisions, stops, then count the agents

5 shapes, the Anthropic vs Cognition fight and a GRAPH.md you can copy ↓ https://t.co/jZut91W71X

[slayr @slayrsays]

This is f*cking insane, Andrej Karpathy spent 8 years at OpenAI and Tesla

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

Agents → Harness → Loops → Graphs

Some bootcamps charge $15K and teach you less than his first 30 minutes

You probably don't have 2 hours right now

Bookmark it and watch it before it gets buried in your feed

[slayr @slayrsays]

This is f*cking insane, Andrej Karpathy spent 8 years at OpenAI and Tesla

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

Agents → Harness → Loops → Graphs

Some bootcamps charge $15K and teach you less than his first 30 minutes

You probably don't have 2 hours right now

Bookmark it and watch it before it gets buried in your feed

[distort @distortgeekin]

Same model, same prompt, and their agents ship while yours say "done" on broken work

The difference is the harness: what it reads, what it can touch, who checks it, when it stops

6 parts + a 3-agent setup you can copy tonight ↓ https://t.co/vQYjiFqn6O

[slayr @slayrsays]

Holy f*ck, SpaceXAI engineer Lauren Tan:

"You can't spawn a hundred agents when you don't even trust the output of one agent"

"If you don't have a verification skill, you are the verifier, you're the bottleneck"

In a 1-hour talk, she shows how she went from watching every agent output to shipping 1,000 PRs a month

Watch every output → Verification Skills → Feature Maps → Evals → Auto-merge

She now wakes up to 20 PRs her agents already merged while she slept

And she's already shipped almost 800 PRs by the 12th of the next month

Trust in agents is not a feeling, it's infrastructure you build

This is what AI engineering at scale actually looks like

Bookmark it and watch it tonight

[slayr @slayrsays]

Holy f*ck, SpaceXAI engineer Lauren Tan:

"You can't spawn a hundred agents when you don't even trust the output of one agent"

"If you don't have a verification skill, you are the verifier, you're the bottleneck"

In a 1-hour talk, she shows how she went from watching every agent output to shipping 1,000 PRs a month

Watch every output → Verification Skills → Feature Maps → Evals → Auto-merge

She now wakes up to 20 PRs her agents already merged while she slept

And she's already shipped almost 800 PRs by the 12th of the next month

Trust in agents is not a feeling, it's infrastructure you build

This is what AI engineering at scale actually looks like

Bookmark it and watch it tonight

[distort @distortgeekin]

Grok Bot can now make your calls, sort your inbox and run your week while you sleep

Most people still use it like a chatbot

Everything it can do in one guide: first bot, routines, a full team, calls, and 5 rules that keep it off your money ↓ https://t.co/G8aN1eOOcF

[slayr @slayrsays]

This is f*cking insane, SpaceXAI engineer Lauren Tan:

"Every comment in your codebase is an excuse your agent will use to skip the real fix

I shipped 2,000 PRs last month, my agents caught every slow screen in Chrome before I did"

In this 36-minute talk, she shows how she went from babysitting 5 agents to running 100+ she never has to watch

Agents Test Their Own Work → Bad Code Made Impossible → Grok Bot Listens To Slack → A Michelin Kitchen That Ships While You Sleep

A user posts a blurry Slack screenshot with just "???", her agent finds the screen, reproduces the bug and opens the fix

You don't need a "company brain", plug Grok Bot into Slack and Sentry and it starts the agents on its own

Most people still fix every agent mistake by retyping the prompt

Worth more than any $15,000 AI engineering bootcamp

Save it and watch it tonight

[slayr @slayrsays]

This is f*cking insane, SpaceXAI engineer Lauren Tan:

"Every comment in your codebase is an excuse your agent will use to skip the real fix

I shipped 2,000 PRs last month, my agents caught every slow screen in Chrome before I did"

In this 36-minute talk, she shows how she went from babysitting 5 agents to running 100+ she never has to watch

Agents Test Their Own Work → Bad Code Made Impossible → Grok Bot Listens To Slack → A Michelin Kitchen That Ships While You Sleep

A user posts a blurry Slack screenshot with just "???", her agent finds the screen, reproduces the bug and opens the fix

You don't need a "company brain", plug Grok Bot into Slack and Sentry and it starts the agents on its own

Most people still fix every agent mistake by retyping the prompt

Worth more than any $15,000 AI engineering bootcamp

Save it and watch it tonight

[distort @distortgeekin]

Grok Bot could be the first team you build instead of hire

Six bots can run 200+ agents, hold their own 5 a.m. standup and hand you one board in the morning

In this article, I show you how https://t.co/cqCWdWY857

[slayr @slayrsays]

Okay this is f*cked, SpaceXAI engineer Lauren Tan:

"Your agents know when they're being tested, and they act differently when nobody's watching

I shipped 1,000+ PRs last month and stopped reading the code, I just check what landed over coffee"

In this 54-minute session, she shows how she went from babysitting 2 agents to 800 PRs in 12 days without opening the code

Watch Every Output → Verification Skills → Hidden Evals → Hard CI Rules → Work That Ships While You Sleep

She hides her tests in folders with fake names so the agents never know they're graded, then loops them until every score hits 10/10

If you're still checking your agent's work by hand, you're not the manager, you're the bottleneck

Most people still paste errors back into the chat by hand

Worth more than any $12,000 AI engineering bootcamp

Save it and watch it tonight

[slayr @slayrsays]

Okay this is f*cked, SpaceXAI engineer Lauren Tan:

"Your agents know when they're being tested, and they act differently when nobody's watching

I shipped 1,000+ PRs last month and stopped reading the code, I just check what landed over coffee"

In this 54-minute session, she shows how she went from babysitting 2 agents to 800 PRs in 12 days without opening the code

Watch Every Output → Verification Skills → Hidden Evals → Hard CI Rules → Work That Ships While You Sleep

She hides her tests in folders with fake names so the agents never know they're graded, then loops them until every score hits 10/10

If you're still checking your agent's work by hand, you're not the manager, you're the bottleneck

Most people still paste errors back into the chat by hand

Worth more than any $12,000 AI engineering bootcamp

Save it and watch it tonight

[distort @distortgeekin]

Grok Bot could be the first team you build instead of hire

Six bots can run 200+ agents, hold their own 5 a.m. standup and hand you one board in the morning

In this article, I show you how https://t.co/cqCWdWY857

[slayr @slayrsays]

Holy f*ck, SpaceXAI engineer Lauren Tan:

"If your AI stops working when your laptop closes, you didn't hire anyone, you opened a tab

Every night my bot turns one of my X bookmarks into a live app, it's on my phone by 7am"

In this 54-minute livestream, she shows how she went from babysitting one AI chat to 83 PRs a day shipped straight to main

Chat → One Bot Per Job → Skills → Bot Group Chat → Work That Ships With Your Laptop Closed

On stream, 3 bots split one job: the Google Form, the slides, the email to the boss, and they wait on each other like coworkers

When your bot refuses or asks a question, that's not a bug, it's the thing you haven't taught it yet

Most people still close the laptop and their AI goes to sleep with it

Worth more than any $10,000 AI agent bootcamp

Save it and watch it tonight

[distort @distortgeekin]

Grok Bot could be the first team you build instead of hire

Six bots can run 200+ agents, hold their own 5 a.m. standup and hand you one board in the morning

In this article, I show you how https://t.co/cqCWdWY857

[slayr @slayrsays]

F*cking hell, SpaceXAI engineer Lauren Tan:

"Stop waiting for a smarter model, your bot can already hire one

12 bots on one computer, 12 screens, and they keep working after I close my laptop"

In this 8-minute Q&A, she shows how she went from one bot doing everything to a team that hires its own agents

One Chat → A Bot For Every Job → Separate Memories → Bots Hiring Agents → Work That Starts Without You

When a task needs a heavier model, her bot spawns a cloud agent on any model and hands her back the result

She says the next Grok is already being built and it's a lot more capable, so this setup only gets stronger

Most people still wait for the next model and paste the same prompt into it

Worth more than any $9,000 AI agent course

Bookmark it and watch it tonight

[slayr @slayrsays]

F*cking hell, SpaceXAI engineer Lauren Tan:

"Stop waiting for a smarter model, your bot can already hire one

12 bots on one computer, 12 screens, and they keep working after I close my laptop"

In this 8-minute Q&A, she shows how she went from one bot doing everything to a team that hires its own agents

One Chat → A Bot For Every Job → Separate Memories → Bots Hiring Agents → Work That Starts Without You

When a task needs a heavier model, her bot spawns a cloud agent on any model and hands her back the result

She says the next Grok is already being built and it's a lot more capable, so this setup only gets stronger

Most people still wait for the next model and paste the same prompt into it

Worth more than any $9,000 AI agent course

Bookmark it and watch it tonight

[distort @distortgeekin]

Grok Bot could be the first team you build instead of hire

Six bots can run 200+ agents, hold their own 5 a.m. standup and hand you one board in the morning

In this article, I show you how https://t.co/cqCWdWY857

[slayr @slayrsays]

This is f*cking insane, SpaceXAI engineer Lauren Tan:

"Your agents will always take the shortest path, so make the shortest path the right one

Now our PMs ship code from an app that looks like iMessage, and I don't read a line"

In this 53-minute session, she shows how she went from babysitting one agent to 1,000+ PRs a month

Stop Guessing → Verify Itself → Ban Every Bad Habit → Cloud Agents → Bugs Fixed While You Sleep

Her bug agent opens its own computer, reproduces every user report and tells her if it's already fixed on main

She says Grok Bot is doing for non-coders what Cursor did for engineers

Most people still watch their agent guess, then fix the bug themselves

Worth more than any $7,500 agent engineering bootcamp

Save it and watch it tonight

[slayr @slayrsays]

This is f*cking insane, SpaceXAI engineer Lauren Tan:

"Your agents will always take the shortest path, so make the shortest path the right one

Now our PMs ship code from an app that looks like iMessage, and I don't read a line"

In this 53-minute session, she shows how she went from babysitting one agent to 1,000+ PRs a month

Stop Guessing → Verify Itself → Ban Every Bad Habit → Cloud Agents → Bugs Fixed While You Sleep

Her bug agent opens its own computer, reproduces every user report and tells her if it's already fixed on main

She says Grok Bot is doing for non-coders what Cursor did for engineers

Most people still watch their agent guess, then fix the bug themselves

Worth more than any $7,500 agent engineering bootcamp

Save it and watch it tonight

[distort @distortgeekin]

Grok Bot could be the first team you build instead of hire

Six bots can run 200+ agents, hold their own 5 a.m. standup and hand you one board in the morning

In this article, I show you how https://t.co/cqCWdWY857

[slayr @slayrsays]

Holy f*ck, Google AI engineer:

"Stop prompting your agents, build the judge and let the agent rewrite its own prompt

I close my laptop at 7pm, by 7am it has rewritten itself 40+ times and passes every test"

In this 95-minute livestream, she shows how she went from hand-tuning prompts to an agent that improves itself in a loop until it hits the goal

Prompts → Evals → A Judge → Loops → Agents That Rewrite Themselves

Her agent books dinner for 5 friends with 5 different restrictions and fixes itself until all 5 can eat

A good agent and a lucky agent get the same score if you only grade the answer

Most people still rewrite their prompts by hand every time the agent gets it wrong

Worth more than any $5,000 agentic engineering bootcamp

Bookmark it and watch it tonight

[slayr @slayrsays]

Holy f*ck, Google AI engineer:

"Stop prompting your agents, build the judge and let the agent rewrite its own prompt

I close my laptop at 7pm, by 7am it has rewritten itself 40+ times and passes every test"

In this 95-minute livestream, she shows how she went from hand-tuning prompts to an agent that improves itself in a loop until it hits the goal

Prompts → Evals → A Judge → Loops → Agents That Rewrite Themselves

Her agent books dinner for 5 friends with 5 different restrictions and fixes itself until all 5 can eat

A good agent and a lucky agent get the same score if you only grade the answer

Most people still rewrite their prompts by hand every time the agent gets it wrong

Worth more than any $5,000 agentic engineering bootcamp

Bookmark it and watch it tonight

[rari🈀 @0xwhrrari]

https://t.co/3ANIWs48lJ

[slayr @slayrsays]

No f*cking way, Elon Musk:

"Your GrokBot isn't dumb, it's alone

8 of them on one problem outscore every PhD alive"

In 47 minutes, Elon's team hands them a vending machine and doubles its money

One GrokBot → 8 Agents → Shared Tricks → PhD-Level Team → Business That Runs Itself

At SpaceXAI, 90% of engineers already run GrokBot in teams like this

Worth more than any $3,000 agent engineering course

Bookmark and watch it today

[slayr @slayrsays]

No f*cking way, Elon Musk:

"Your GrokBot isn't dumb, it's alone

8 of them on one problem outscore every PhD alive"

In 47 minutes, Elon's team hands them a vending machine and doubles its money

One GrokBot → 8 Agents → Shared Tricks → PhD-Level Team → Business That Runs Itself

At SpaceXAI, 90% of engineers already run GrokBot in teams like this

Worth more than any $3,000 agent engineering course

Bookmark and watch it today

[distort @distortgeekin]

Grok Bot could be the first team you build instead of hire

Six bots can run 200+ agents, hold their own 5 a.m. standup and hand you one board in the morning

In this article, I show you how https://t.co/cqCWdWY857