May 2026

Time for another monthly post sharing some of the IT discoveries that crossed my path lately. Below you'll find a mix of articles, links, and resources, some of which tie into my current activities and areas of interest.

Some thoughts

Half Quill

My colleague created Half Quill, a collaborative writing platform. Its core concept is "picking up where a stranger left off." If you want to exercise your brain, practice your writing skills, and prevent your brain from rotting, this might be the way to go. There is also no pressure to write, you can just read and see where other people's minds drifted off to. I also noticed that their latest update allows you to download stories as ebooks. It just keeps getting better and better.

Cursor Meetup Amsterdam

I had the pleasure of attending the Cursor Meetup at the Booking headquarters in Amsterdam. Visiting the Booking office was an adventure in itself. I didn't get to see all of it, but it was still pretty impressive. There were a few presentations; some dove deep into AI technicalities and applications, while others were closer to the day-to-day developer experience. I especially enjoyed two of them, both given by guys who also work at Booking.

Roy Klein — Engineering Discipline Now, More Than Ever.

I really liked his slide featuring a massive "10x dev, 100x maintenance" quote. With great power comes great responsibility. AI lets you produce code faster, but the real question is how to do so without compromising quality and sustainability. The short TL;DR is that you should batch your changes, and your CI pipeline is the deciding factor in how small those batches can be (if your CI takes 30 minutes, it might not be a great fit for one-liners, but if it takes 2 minutes, your batches can be much smaller). So, the goal is fast CI, paired with a strong emphasis on refactoring. I can't argue with these points; it sounds perfectly reasonable.

Gal Naor - Don't Trust The Code You Didn't Plan

I read somewhere recently that productivity doesn't matter if you work on the wrong thing. With the rapid pace of code generation today, planning has become more important than ever. Gal presented his SDLC plugin for Claude Code and Cursor, which makes code generation much more intentional by using planning as a comprehensive gatekeeper. It helps you answer crucial questions upfront: WHY, WHAT, and HOW.

github.com/galando/temper

Finally, I really enjoyed the Cursor team's talk on the future of AI-driven engineering and their vision for the product (TL;DR: Autonomous Development and Intent-Oriented Product Creation).

Roy Klein — Engineering Discipline Now, More Than Ever

Gal Naor - Don't Trust The Code You Didn't Plan

Tooling

I've been exploring some tools lately. Here are some highlights:

hurl - tool for running HTTP requests defined in plain text format
sidebery - Firefox extension for tab management
Unhook - YouTube extension to remove distractions

Articles

Learnings from conducting ~1,000 interviews at Amazon

Steve Huynh, formerly Principal Engineer at Amazon, shares observations from 10+ years of interviewing software engineers, and an excerpt from his new book, Technical Behavioral Interview

Why TUIs are back by Alcides Fonseca

AI generated summary

The article explores the comeback of Terminal User Interfaces (TUIs) due to the challenges and inconsistencies in modern graphical user interfaces (GUIs) across various operating systems. It critiques the fragmented state of GUI development and advocates for a return to fundamental UI design principles to achieve consistent user experiences. wh

Small Programming Tricks · will keleher

A lot of effective engineering comes from the accumulation of small nuggets of knowledge.

Git Merge Conflicts: Understanding Ours, Theirs, and Base | Code Input

Merge conflicts are less complex once you understand the three versions git is comparing: ours, theirs, and base. This article walks through how conflicts happen, what the conflict markers mean, and how to find the commit that introduced a conflict.

Tyblog | You Don't Love systemd Timers Enough

A cron job for every man, woman, and child.

Serving a Website on a Raspberry Pi Zero Running Entirely in RAM

AI generated summary

It covers the necessary hardware and software setup, including the use of a VPS for TLS termination, and offers tips for maintaining persistent configurations and backups.

A Rare, Data-Driven Look at the True Cost of API Integrations – Globick

AI generated summary

This article explores the comprehensive costs of API integrations, emphasizing that while initial development is costly, ongoing maintenance and operational expenses are significant. It highlights the underestimated nature of these costs and discusses the limited impact of standardization and AI in reducing them.

Should I Run Plain Docker Compose in Production in 2026? | Distr

Yes, plain Docker Compose can still run production workloads in 2026—if you close the operational gaps it leaves: cleanup, healing, image pinning, socket security, and updates.

Why are the Artemis II photos on Flickr? - Anil Dash

A blog about making culture. Since 1999.

Productivity

Finishing Things

This was originally supposed to be a pretty short note. But it turned out I had a bunch of related thoughts all tangled up. ... What follows is a bracingly honest assessment of how I do (and don’t) sometimes finish projects.

AI

Appearing Productive in The Workplace — No One's Happy

AI can produce work that looks expert without being expert. The failure arrives in two shapes, and both are reshaping the workplace.

My AI Workflow (Without Losing My Skills)

AI generated summary

The author shares their AI workflow, focusing on balancing AI-assisted coding with manual coding to retain skills. They discuss the risks of skill erosion due to automation and provide a detailed example of updating an app using a mixed approach.

Text is Thought, and Thought is Holy | Daniel Miessler

Why I think Markdown is a better spec format than HTML

Agentic Coding is a Trap | Lars Faye

Remaining vigilant about cognitive debt and atrophy.

The Pulse: ‘Tokenmaxxing’ as a weird new trend - The Pragmatic Engineer

At Meta, Microsoft, Salesforce and other large companies, devs are purposefully burning tokens (and money!) to inflate their AI usage and hit AI usage metrics which they treat as targets.

The last six months in LLMs in five minutes

Simon Willison's blog post summarizes the developments in large language models (LLMs) over the past six months, as presented in his lightning talk at PyCon US 2026.

What I Learned Building Multi-Agent Systems from Scratch - InfoQ

Paulo Arruda shares Shopify’s journey from "vibe coding" to building a multi-agent microservices architecture, exploring how specialized AI agents and context engineering maximize engineering ROI.

The AI Bubble — No One's Happy

The AI buildout is the largest capital expenditure in the history of the technology industry. The financial structure holding it together has a name.

DHH’s new way of writing code - by Gergely Orosz

David Heinemeier Hansson shares why he shifted to an agent-first AI workflow, and what it means for how software is built and who builds it.

GitHub - addyosmani/agent-skills

Production-grade engineering skills for AI coding agents.

Python

The Simplest MCP Example Possible in Python - Invent with Python

AI generated summary

This page explains how to set up and run a local LLM using the Model Context Protocol (MCP) to access tools for getting the current time and date. It includes instructions for installing necessary packages and running scripts, as well as discussing potential issues with LLMs.

Software development training - Geekuni blog: Learn concurrency - a deep dive into multithreading with Python

This article explains concurrency in Python including topics like multithreading, multiprocessing, race conditions, and synchronization mechanisms such as locks.

Django

Djangocon EU: when SaaS is not allowed: shipping Django as a desktop app - Jochen Wersdörfer - Reinout van Rees

AI generated summary

Jochen Wersdörfer presented at Djangocon EU on deploying Django as a desktop app using Electron, due to security constraints preventing SaaS deployment. The solution involves Electron, Django, and SQLite, with security measures and packaging challenges addressed. Alternatives and related talks are also discussed.

Tools

Lit

Simple. Fast. Web Components.

Unhook

Unhook YouTube is a tool designed to enhance the YouTube viewing experience by removing distractions. It allows users to block suggestion feeds, comments, and other potentially distracting elements on YouTube.

GitHub - Orange-OpenSource/hurl

Hurl, run and test HTTP requests with plain text.

GitHub - httpie/cli

🥧 HTTPie CLI — modern, user-friendly command-line HTTP client for the API era. JSON support, colors, sessions, downloads, plugins & more.

GitHub - open-saas-directory/awesome-native-macosx-apps

The best Mac apps — fast, lightweight, and bloat-free. No Electron. Curated for Mac power users who care about performance.

daisyUI - Tailwind CSS Component Library

Tailwind CSS component library by daisyUI. Build faster with semantic components, built-in themes, and reusable UI blocks.

GitHub - mbnuqw/sidebery

Firefox extension for managing tabs and bookmarks in sidebar.

Cloud

Basecamp — Leaving the Cloud — Cloud Computing Isn’t For Everyone

We’ve run extensively in both Amazon’s cloud and Google’s cloud, but the savings never materialized. So we’ve left.

Other stuff

Half Quill

Pick up where a stranger left off

A five-minute daily writing ritual and a community where stories grow one section at a time. Every word is human.

Plain Text Accounting - plaintextaccounting.org

Plain text accounting is a way of doing bookkeeping and accounting with plain text files and efficient, command-line-friendly software like Ledger, hledger, or Beancount.