Simão Belchior

Writing

Notes, now and then.

About work, what I'm reading and whatever else catches my attention. Some posts are in Portuguese.

Subscribe via RSS

  1. October '26 Stars

    · 1 min read

    This month brought a shorter list, though I feel like I might've forgotten to star something... but here it is.

    1. Panache — RTS Belle Époque pour mobile by Gabriel Turpin (via Simon Willison)
    2. How abliterated models can get you pwned by Prince Chaddha (via Bruce Schneier)
    3. Engenharia Reversa com IA: a Adobe morreu? by Fabio Akita
    4. Claude, don’t touch my lump by Adam Mastroianni
    5. Auto-publiquei meu próprio Livro/e-book: Competente o Suficiente by Fabio Akita
    6. Existem Linguagens Melhores pra programar com LLMs? by Fabio Akita
    7. We are going to kill “unalive” by Anil Dash
    8. The Code Nobody Reads by Addy Osmani
    9. Bliki: Sensible Default by Martin Fowler
    10. I Want Better Reporting on AI Genie Behavior by Bruce Schneier

  3. September '26 Stars

    · 2 min read

    It's been a while since I shared my reading list on the blog, but here goes my starred articles from September.

    1. Note on 24th September 2026 by Simon Willison
    2. What About Rails? by Jared Norman
    3. Tyrannosaurus by Randall Munroe
    4. Wake up, people. What we should actually fear, near term, is not so much rogue superintelligence as unleashed agentic AI causing hacking the internet at scale. by Gary Marcus
    5. How To Write With An LLM by Thomas Ptacek (via Simon Willison)
    6. I don't like LLMs by Martin Fowler
    7. Two cheers (out of three) for Dario Amodei by Gary Marcus
    8. Out-Of-Distribution Checks – How Human Intelligence Stabilises Agentic Workflows by codemanship
    9. A.I. Was Supposed to Give Us New Killer Apps. What Happened? by Paul Ford (via Simon Willison)
    10. Could rogue agent swarms take over the entire internet in the next six months? by Gary Marcus
    11. P(doom) by Armin Ronacher
    12. Feeling sad about AI by Andy Balaam (via Simon Willison)
    13. No, Anderson Cooper, AI is not going to kill all humans by 2030 by Gary Marcus
    14. Você é um idiota se acredita nas propagandas enganosas da OpenAI, Anthropic, NVIDIA. Entenda by Fabio Akita
    15. What We Can Learn from Claude's Fable 5.1 System Prompt by Drew Breunig
    16. The Ultimate Equation about AI Coding by codemanship
    17. I'm done using AI by brettcodes
    18. Agentic Horizons – When The Wheels Start To Wobble… by codemanship

    Most is about AI since that's most of what I read these days, maybe I should expand my feed...

    Continue reading →

  6. 🥷 Stealth

    · 1 min read

    One thing that AI seems to have brought into this world is the concept of "building in stealth mode"... maybe it's because AI makes it easier to create greenfield projects and that helps having a go at creating a new startup...

    Or maybe people are so used to posting on Social Media that when they go a few months without telling everyone what they are working on, they feel like they need to call it stealth... because it's fancier.

    Who knows really...

    stealth building on Google Trends

  7. January '26 Stars

    · 2 min read

    January always felt like a long month, but as I grow older and time becomes shorter it almost feels like January keeps growing. Or at least it's length is more noticeable. It might also just be because it was a particularly wet January... Who really knows?...

    Whatever how it felt it was a month of much reading, from the 65 feeds that I follow I read 221 articles and starred 26 of those. You can find those favourites below.

    1. We are all doing AI wrong by nutanc
    2. Moltbook is the most interesting place on the internet right now by Simon Willison
    3. Singing the gospel of collective efficacy by Matt @ Interconnected
    4. Productivity Theatre @ codemanship
    5. The Five Levels: from Spicy Autocomplete to the Dark Factory by Simon Willison
    6. The 80% Problem in Agentic Coding by Addy Osmani
    7. Claude's new constitution by Simon Willison
    8. Giving University Exams in the Age of Chatbots by Simon Willison
    9. Agent Psychosis: Are We Going Insane? by Armin Ronacher
    10. Bridges by Kent Beck, I talked about this article on a blogpost in this blog.
    11. I Have Thus I Deserve from Dark Thoughts
    12. AI Agents: Garantindo a Proteção do seu Sistema by Akita on the importance of security when using AI coding agents (PT text!)
    13. A Software Library with No Code by Drew Breunig
    14. LLM predictions for 2026, shared with Oxide and Friends by Simon Willison
    15. When AI writes almost all code, what happens to software engineering? by Gergely Orosz. We covered this on one of our AI Open Hours @ Mindera as it's a really hot topic and with a lot of uncertainty of what's to come.
    16. The big regression by Jason Fried on how design has
    17. Welcome to Gas Town by Steve Yegge, introducing an interesting take on multi-agent coding via an AI Orchestrator. Pretty wild.
    18. The Future of Coding Agents by Steve Yegge
    19. Introducing gisthost.github.io by Simon Willison
       
    Continue reading →

  8. On building bridges

    · 1 min read

    I just read this article about building connections and bridges with others, written by Kent Beck. I received it via email from a Substack newsletter, and got hooked by the intro. 

    It's so sad that intro, the story. It's one of the biggest fears any parent carries, and reading about it really moved me. But what made me share it was what came next, the path of learning how to build connections. It's a really interesting read for me. Maybe because it's another example of the power of learning.

    There's a Portuguese saying: "ninguém nasce ensinado", which can be translated to "No one is born knowing everything.". This is very obviously true in regards to technical things, but it's important to know that it is also true for other things... like connecting with others.

    There are lots of articles out there about how humans are becoming lonelier... but fear not, because there's hope. :)

  10. Time in the Zone

    · 2 min read

    Mindera is a real global company, with offices in 9 countries and people working from multiple other locations.
    When my team started organising our AI Open Hours we decided to communicate the time of that meeting in all the locations, to make it easier for everyone to figure out if it would work for them. That meant that our messages were always that much longer and noisier.

    Today, as I was preparing another message I decided to find a tool that would allow me to share just a single time in my current timezone and have people easily figuring how it would translate to their location. After a couple of google searches I gave up on that path and decided I'd just use Github Copilot to create a simple tool that would give me what I wanted. And so Time in the Zone came to be.

    It's a one page tool built with HTML, Javascript and CSS that shows the current time in Portugal as well as in other Mindera offices' locations and other remote timezones. Besides that it also allows us to see times for a specific time, date and even changing the reference location, all using query parameters.

    The configurable parameters are:

    Given that this was a tool with a small and clear scope, and which won't have any further dependencies or with any associated risks it was completely vibe coded, using Github Copilot. I provided the ideas and the features definitions and then asked Copilot built it without me paying any attention to the generated code, only to the generated features and functionality, ensuring that it did what I wanted it to do. Or adjusting when it wasn't quite right. This is quite a big difference from when I'm using AI assistants to build production level applications. In those scenarios all AI generated code needs to go through careful code reviews to ensure best practices, consistency and high quality outputs.

    Continue reading →