AI recommended a luxury hotel that had been demolished 108 days earlier.
Happy Halloween: every hotel has a ghost, and AI keeps sending guests to meet it.
What's actually real about booking with AI right now (and what isn't):
hotelemarketer.com/2026/10/3…
You have two jobs.
Only one is in your job description.
That's the one AI is about to take.
The other is why you'll still matter: hotelemarketer.com/2026/10/0…
Opus 5.5 can't generate video. So I asked it to write one.
Hotel marketing vs 20,000 real guests, as an 8-bit fighting game. Final boss: "Luxurious."
Built on yesterday's Opus + Jev experiment.
hotelemarketer.com/2026/09/2…
Opus 5.5 + Jev is a ridiculous combo for high-volume judgment work.
I fact-checked 15 hotel marketing clichés against 20,000 guest reviews.
640,000 typed judgments. 11 minutes. $3.74.
Opus doing it alone: ~$230–490. 🧵
hotelemarketer.com/2026/09/2…
We pay for five streaming services and still can't decide what to watch.
So I taught an AI my taste from 900 of my own ratings. The twist: the clever prediction barely beat a coin toss. The boring stuff won.
Story + demos: hotelemarketer.com/2026/09/2…
Everyone's panicking that AI is slipping out of human control.
Meanwhile, a man who lost his voice to ALS just spoke to his wife again through a brain implant.
Same tech, two directions. We're still driving. 👇
hotelemarketer.com/2026/09/1…
Two of the best AI models. One 30-second film.
Same brief: skydive in on a deadly-serious mission to deliver… a plastic beach spade.
Claude vs ChatGPT, each directing its own cut. Two very different films, back to back.
Which one would you hire?
hotelemarketer.com/2026/09/1…
Two weeks of pain. Four dentists. One health scare.
The AI I leaned on never diagnosed a thing. It just quietly held the thread between every appointment.
Here's what that actually looks like: hotelemarketer.com/2026/08/2…
I can't write a line of code. Last month I built two working apps anyway.
One's a space game where you run a hotel on an orbital station. The other reads my Oura ring to help hunt down my morning headaches.
hotelemarketer.com/2026/08/0…
"As soon as it works, no one calls it AI anymore."
A friend builds AI agents he talks to like colleagues. One filed 30 invoices over lunch. Every one was fake. The reason why is the best AI reality-check I've read this year.
Full conversation here - hotelemarketer.com/2026/07/1…
AI just made "impressive" free.
Anyone can now produce brilliant-looking work on any topic before the kettle boils.
Which means the whole game is now judgment: can you tell whether the brilliant-looking thing in front of you is actually any good?
hotelemarketer.com/2026/07/1…
When AI can make anything, the scarce skill is taste, knowing which moment matters.
The people winning with these tools aren't the fastest. They're the best at judging "not that, this."
That's the 2 a.m. test. Full story + film: hotelemarketer.com/2026/06/2…
My kids think AI is sick. They're right. That's exactly what worries me.
So I gave them the AI talk, boiled down to something they won't forget: U SICK.
Made it a free comic you can download and share 👇 hotelemarketer.com/2026/05/3…
8pm on Friday. 11 slides into a 30-slide deck. My AI just ran out of fuel.
Wait 4 hours, or pay extra. I paid. Of course I paid.
For the first time in my life, I could watch the cost of thinking tick down in real time.
That's how token anxiety starts.
hotelemarketer.com/2026/05/2…
We are getting very close to editing video by simply talking to it.
I tested Google’s new Gemini Omni Flash this weekend.
People into puppets.
Car into rhino.
My avatar into a video scene.
Colour changes + sound effects.
Not perfect. Not cheap.
But wow, this is coming fast.
A better travel prompt is fine. A better workflow is the unlock.
Just published mine. End to end. Includes the destination prompt I copy-paste before every trip. Yours to steal:
hotelemarketer.com/2026/05/2…
Aluminum was once more precious than gold.
We capped the Washington Monument with it.
Within decades, we wrapped sandwiches in it.
PhD-level expertise is next.
What that means, in plain English →
hotelemarketer.com/2026/05/0…
An AI made the warning about AI.
81 seconds on why your chatbot isn't your friend, made by someone who can't animate, in one Saturday afternoon. Worth a watch before your kid's next conversation with one.
Full story: hotelemarketer.com/2026/05/0…
ChatGPT Images 2.0 can do way more than people think.
Magazine covers. Movie posters. Your own cereal brand. Comic books. Interior design. 360° images you can look around inside.
10 experiments. Prompts included so you can try them yourself.
hotelemarketer.com/2026/04/2…
I asked an AI to draw a walking map from the Ritz Paris to the Louvre.
The street names were real. The landmarks were in the right places. It said BONNE PROMENADE! in the corner.
ChatGPT Images 2.0 dropped yesterday. I ran 7 hotel-specific experiments:
hotelemarketer.com/2026/04/2…


