Turning Pre-Arrival Engagement into Measurable Revenue. I recently delivered a keynote at the SPARK Start-Up Summit at @LesRochesNews. My conversations with luxury hospitality operators kept circling back to one critical question: How do we build value in a moment where teams are overworked? The modern luxury guest expects a flawless experience, but the window to shape that starts 4–6 weeks before arrival, not at check-in. That’s where Vibeslist comes in: 1. See exactly what is happening in the city during your guest's stay. 2. Share a curated digital event list before they arrive. 3. Extend the stay. Guests who plan ahead add +0.3 nights to their average length of stay. After 18 months of development, our product is ready for enterprise-level delivery, and our first hotel partners are already live. Time to speak with more GMs.

Everyone: "Oh wow, the Eiffel Tower! So majestic." Adorable. Startup founder? Taking notes on the OG moat building.

I’ll be honest... I was hesitant. When you have a ruthless quality standard for your data pipelines, you don't just change things overnight. But after 5 days of obsessive monitoring... we did the unthinkable. We ripped out 70% of our infrastructure and handed it to Luna. 🌕 The result? Quality didn't drop. It actually went up. But here is the part that is going to blow your mind: our costs dropped by 90%. Ninety. Percent. 📉 When tokens become this cheap, it completely rewires how you think about architecture. 🧠 Building Events Intelligence just got even more fun. 🔥

Next week's itineraries for San Francisco, Milan, Florence, Rome, and London are now deployed to our hotels partners. The attached previews represent a fraction of the full intelligence briefs operating inside partner properties. Pre-arrival teams now hold the exact insights, from private caviar dinners and open-air opera to exclusive film restorations, required to solve the itinerary bottleneck before guests check in. @vibeslistai is the enterprise platform turning pre-arrival itineraries into measurable revenue. Set the standard.

I put our latest AI hotel agents to the test and generated this clean Memorial Day Weekend itinerary for SF.

Code refactoring and good architecture is something that we all learnt by making thousands of mistakes and manually rewire and polish our own code. A simplification there, a over engineering there, a silly best practice there. But we made the decision and we took ownership of our own code. Now there is a very deep philosophical question, if you are robbed of this learning opportunity as Vibe Coding just gets the job done, how the heck you will learn? Like for me Vibe Coding is great because I have 20 years of coding experience and I made so many mistakes I just know when something feels off and i need to push Codex or Gemini into a different direction. But if you don't have that manual experience, how do you know when that is not right? Sure other people reviewing, sure some code reviewing tools, but still. Building software is at more an Art that just pure Science. I don't have an answer, just very curios to see how the new Junior Devs will learn the Art of great coding and great architecture if they never write code by hand.

Sometimes I wonder why they don't teach how to Code Review a PR in college. I mean, I understand there is a lot of believe cooked in what makes a good or bad PR, but I feel like just some foundational work like: - when is worth add comment - must this have tests - what makes a test a good one - is this code a potential performance hazzard - is this code easy to mantain - when is ok to push the limits and when is not Maybe because it gets into the Art territory instead of Science, it can get hard to validate. But I think helping young software engineer practice more the soft-skills in how to think good vs bad code, it's so critical, ESPECIALLY in the Vibe Coding Era.

Apperently Gemini is actually a better experince, today, Dec 9th 2025 than ChatGPT. I honestly did not believe that, but in the last 36 hours I end up having a much better experience with Gemini. Ok, so what's actually happening. The feeling I have is that ChatGPT is very very Academic kind of thinking, it gives you what you need but it's more School smart. But what I need most of the time is Street Smart. I want an AI that knows how to get the job done, don't talk too much, be able to anticipate my thinking, and cut the conversation when need it. The race for the best Consumer AI is for sure like F1 2025 season, very very close, and you should never fully count out Max. Well Done Gemini! And can't wait to see how this race will play out in 2026.

The Christmas vibes were just right 🎄@theallinpod 🎄

I refactored so many projects in the last 20years of coding, that as CEO I end up having my team pushing at max speed every week, and then I stop by the codebase for an afternoon, look their code and do a big big big cleanup. Maybe because I spent so many years built Infra, that I just can not have chaos in our codebase. Or actually let me reframe it. Chaos is good, but I still want to make sure we can be productive. The last thing you want in any codebase is when chaos ends up slowdown productivity by 2~3x. One of my favorite memories when I was 22 was rewriting the same CMS I built from Php to Python to Node.js, and from SQL, to Postgress to Mongo. I know, why? But hey, that was the fun of loving the craft. To be fair, once I was so mad at jQuery I end up just writing my own jQuery-Like library called Tiramisu, with @proudlygeek . Best part of it, was that instead of calling $.wathever we had t.whatever aahaha @proudlygeek made a sick landing page, I still think about it time to time how good it was. it's is still up, which is crazy opentaste.github.io/tiramisu…

When you build an Infra that can build clean dataset, oh boy everything gets so easier to build on top of it. Then the real question is, what's good Infra in the age of AI? Honestly I have my own thinking around it, and it change the answear based on what phase of your product you are. Internally in Vibelist we started using the term Gen2, Gen3 as a way to define clear chapter of the different generation of Infra and Dataset we are building. Gen 1 was very chaotic, honestly we had good intuition but not good structure, Gen 2 we really were able to build solid taxonomy, and create several internal tools to generate the labeling and vector we need. So no we just finish Gen 2, literally today. Over 2.5k complex record all well classified and 20+ Taxonomy to support it. Gen 3 which will last probably till end of Janaury is focus a lot on refining how the this structure needs to get more granular and let's just say smaller pieces of legos. If you are in SF and love geek around Taste and Dataset, I am around to chat over incredible coffee.

I feel like more and more X is LinkedIn and Linkedin is X

🧢🌁

🎂 Vibes

wait is Wednesday? xitter.cf/buildItN0w/status/1993…

Buona Santa Cecilia 🎄

X should have Badge like Vibeslist has just some way to tell a Post has the Vibe you are looking

Just honest question How do I get X to stop showing engagement farm posts?

Walking is dangerous for a visual founder I keep getting more ideas on how to improve the product