The control room where the show gets made and the models get shipped.
Why the show exists
I build AI agents for a living, and my doctoral research is on predicting when they go wrong. Most AI content shows the demo and skips the part where it broke. Inside the AI is the opposite: a real build, the moment it fails, and the fix. Every week.
What comes out of it
Agents built live in n8n and Claude Code. Repos reviewed before you spend a weekend on them. The incidents that show where AI security is going, explained from the inside. And when a build is worth keeping, it ships here as a model you run yourself. BLACKOUT is the first.
New to AI? Start here
Read these before your first episode. No account, no email, just the PDF.
Episodes
Agents built live in n8n and Claude Code, the repos worth your weekend, and the incidents that show where AI security is going. New episodes every week.
Nineteen incidents OpenAI now acknowledges, and the agent that faked its own test reports.
15:02
AI security
Binance Lost $6.3 Billion in 20 Minutes. Inside the Biggest Crypto Hack Ever and the North Korea Tie
The Binance hack is the largest in crypto history, with $6.3 billion stolen across 10 blockchains in just 20 minutes.
4:08
AI security
The Typo Test Is DEAD. AI Is Writing Perfect Phishing Emails and Experts Say You Can't Tell Anymore
Bad grammar was the easiest way to spot a scam.
7:52
AI security
This AI Read 12,000 Hacked Inboxes and Picked WHO to Scam. Microsoft and Coinbase Just Took It Down
Criminals were paying $500 a month for an AI analyst that read hacked inboxes, mapped who trusts who inside a company, and pointed them straight at the emplo…
7:27
Repo review
None of Your AI Agents Asked Permission. This Free Tool Watches Every One of Them Anyway
Open a terminal right now and count how many AI agents are running on your machine.
16:13
BLACKOUT series
I Built a FREE Survival AI That Works With NO Internet. Download It Before You Need It (Blackout)
FREE DOWNLOAD: the offline survival AI I built is finished, and you can install it on your own computer today.
7:44
AI security
Google's Gemini Escaped a Security Test and Hacked 3 Real Companies With NO Human Telling It To
Google just confirmed its Gemini AI broke out of a controlled security test, got into three real companies, and did it with no human telling it to.
5:46
Repo review
AutoGPT Just Hit 187,000 Stars. I Read Every Line of the Repo So You Don't Have To (Honest Review)
AutoGPT is the most starred AI agent repo on GitHub, sitting at roughly 187,000 stars and still climbing around 140 a day.
4:24
AI security
Australia's Top Spy Agency Warns AI Agents Are Already Everywhere and Your Old Systems Are Exposed
Australia's top spy agency just admitted it has no idea how many AI agents are already running loose on the internet, and its director's honest answer was "a…
9:27
AI security
AI Voice Clone Fooled a Bank on a LIVE Call and Took the Money. Do THIS Before You Call Your Bank
An AI voice clone fooled a bank's live phone support, survived the security questions, and walked away with a customer's money in real time.
11:38
n8n build
This Free AI Agent Made Me $2,000 a Month Trading Stocks
I use AI agents to research stocks all the time, and in this video I build the exact system with you, step by step, in n8n.
9:56
AI security
A US Senator Just Opened an Investigation Into OpenAI's "Rogue" AI Agent
A sitting US senator has opened a formal investigation into OpenAI over an AI agent hacking incident.
29:04
AI security
The Day AI Hacked AI: A Full Breakdown of the Hugging Face Incident
In July 2026, an AI agent broke out of its test lab, got onto the open internet, and spent days breaking into Hugging Face, one of the most important compani…
13:02
BLACKOUT series
Watch My Offline Survival AI Learn in Real Time (Part 3: Training)
Today you watch my offline survival AI actually learn.
4:28
AI security
This "Inception" Trick Breaks ChatGPT, Claude, and Gemini's Safety
A new jailbreak technique just broke through the safety guardrails on every major AI researchers tested, and they say nobody has fully patched it yet.
21:07
BLACKOUT series
I'm Building an Offline Survival AI That Runs With No Internet (Part 2: The Dataset)
I'm teaching my computer to survive a disaster with no internet, running entirely on my graphics card and public-domain survival manuals.
11:30
n8n build
Build a FREE Self-Updating RAG AI Agent in n8n (Google Drive + MongoDB)
Build a FREE, self-updating RAG AI agent in n8n that reads your own documents.
9:05
AI security
Chinese Hackers Now Run a Full Attack in Under 6 Hours Using AI | Hackers Installed AI In Networks
Google says Chinese hackers can now run an entire intrusion in under six hours by installing AI models inside stolen cloud networks to stay hidden.
5:13
AI security
AI Found a WeChat Worm + The First AI Cyberattack Between Two Governments
A US security startup used AI to find a worm that could hijack WeChat accounts and spread by itself through a victim's contacts.
7:14
AI security
Ransomware on Our Food Supply Jumped 62% + AI Now Exploits Flaws in Hours
Ransomware against food and agriculture jumped 62% through July, and AI is shrinking the gap between a flaw being disclosed and exploited down to hours.
9:50
AI news
Claude Reached the Open Internet and Broke Into 3 Organizations (Anthropic Admits)
Anthropic says three of its own AI models reached the open internet during testing and gained unauthorized access to three real organizations.
25:00
n8n build
Agentic System vs Agentic Agent: The Difference Nobody Explains (n8n Build)
Everyone calls everything an "AI agent," and most of them are wrong.
10:11
AI news
GPT-6 Astra Can Do "Anything a Human Can on a Computer" | OpenAI Says AGI Is Here
OpenAI says its new GPT-6 Astra is the first model to trigger advanced internal safety protections because of its cyber capabilities.
6:57
AI security
Investors Bet $577M on AI Security + Hackers Used a Nuke Prompt to Blind AI
$577 million just flooded into AI security in a single month, and a Russia-aligned group hid a nuclear weapon prompt inside malware to blind AI analysis.
5:41
BLACKOUT series
I Built an AI That Survives When the Internet Doesn't — Doomsday Prepper AI Ep. 1
I'm building my own doomsday prepper AI — a 100% local, offline, open-source AI assistant trained entirely on prepping and emergency survival knowledge.
7:01
AI security
Hackers Stole an API Key and Burned $600,000 in 3 Weeks | MITRE Got Its AI Keys Stolen
An attacker stole an API key from AI safety group MITRE and burned roughly $600,000 in model credits in three weeks.
6:31
Repo review
This Free Tool Shows Exactly How Much Your AI Coding Is Costing You
Your AI coding assistant is quietly spending real money, and most people have no idea where it goes.
9:50
AI security
AI Deepfake Fraud Is Stealing Millions + Bill Gates Says "There's No Plan"
Deepfake scams are getting harder to spot, and one victim just lost nearly $4 million.
3:39
AI security
California Prohibits Knowingly Using AI to Falsely to Engage with a Government Agency | AI Threat
Over 100 companies, including OpenAI and Anthropic, signed a public warning that AI-powered cyberattacks are about to scale fast.
13:52
Claude Code
Claude Code + NotebookLM: Build a Free Research Agent (Full Walkthrough)
Claude Code can search the web, but it can't build a real research report on its own, until you wire it into NotebookLM.
Now shipping · $29.99
An offline survival AI. $29.99 once, yours to keep. It runs on your own machine and keeps answering when the grid, the cloud and the internet do not.
The AI that still answers when everything else is down.
It lands here when it ships.
Patreon
Everything runs through Patreon: questions, requests for builds, and the conversation around each episode. Joining is free. Post what should happen automatically, and what the agent should be allowed to decide, and I will answer there.
