Retrospectiva #4
Well—that was quick. Just like that, the first month of the year is gone. Winter blues would normally peak around this time. But not this year. We're spending time with family in my favorite place on Earth, somewhere along the Adriatic coast of Italy, in the Marche region.
It's hard to give into winter blues when you get to run every day around the beautiful hills of the Conero Regional Park. If you read this, don't tell anyone please: let them visit the usual suspects instead.
Here, life moves slow. You enjoy things. You savor things. You talk loud. You socialize. You live for 100 years. It's Italy.
Using
I would be lying if I told you I haven't been using a bunch of agents. I've been using LLM based tools for the better part of 2025. But something clicked towards the end, and I haven't looked back since. As for my post on policing agents, my fear is still the same. That quantity and quality are inversely proportional. All I can do is find out ways of keeping quality high. It's nice to fire them off to do work I'm not interested in doing. Like redesigning this website for the Nth time. Or correcting typos in all my blog posts since the beginning of time.
On another note. We always travel light, even with a newborn. This means I have to pack just enough things - especially running gear. The FuelCell Rebel v5 have been a pleasant surprise. I'm not loyal to brands but this model won me over. They slip where it's muddy though, so beware.
Reading
Read more short form than long form this month. I continue going through the fabulous long form writing from the Huggingface team. Lots of small tips and tricks, and scars on how to train Large Language models. Elon is a divisive character, but one thing I do appreciate is the opportunity of reading through the recommendation algorithm that powers X.com in my free time. Finally, the GPTZero folks have written this great "GPTZero finds 100 new hallucinations in NeurIPS 2025 accepted papers" piece about hallucinated references in NeurIPS papers. The world is getting filled with AI generated content.
I'm still finishing up both books I talked about in the last issue: Old Man's War from John Scalzi and LLMOps.
Listening
I follow a lot of DJs on SoundCloud. One that has consistently released good mixes is Sasha Marie. It's the best soundtrack for doing deep work. Whatever it is you're doing, you'll do it better when this is in the background:
Two other recommendations that have been on repeat this month. A bit of modern, and a bit of smooth.
On another note, two podcast recommendations this month:
- This interview from The Knowledge Project Podcast with Morgan Housel: I love how Morgan thinks, talks, and writes. His books are gold - and no wonder he's sold so many. He just resonates.
- State of Agentic Coding with Armin and Ben: Armin Ronacher is one of those people at the bleeding edge of agentic workflows. Always interesting to understand how he's thinking.
Watching
While in Portugal, Vitto and I binged The Pitt on HBO Max. It's like Grey's Anatomy, on steroids, at 4x speed, and 5x the intensity, with a lot less drama. Or at least the soap opera one. We haven't finished it, but definitely recommend it.
Since we're in Italy, and it's Australian Open time, we've been watching a good amount of that as well. Tennis skyrocketed here in popularity due to Sinner, Musetti, and Paolini. So when Tennis is being played, Italy stops. But it's not only sports. We also watch intellectually rewarding things. But not this month.