All Articles
Discover all our articles about development and AI
Vibecoding and security: the hidden risks of coding with AI
Vibecoding accelerates development, but what are the security risks that AI silently introduces into your code?
There are two types of developers, and AI is widening the gap
AI divides developers into two camps: those who code for the intellectual challenge and those who code to bring a project to life. Pick your side.
The technology gap: between revolutionary AI and broken HTML forms
We can prototype apps in days thanks to AI, but my bank still can't make a form work. Customer support is going to be a massive market.
What is MCP (Model Context Protocol)? Hands-on experience with Cursor
MCPs will revolutionize AI usage by connecting all useful programs to it. It's not mature yet, but some implementations are already impressive.
Configuring Cursor with custom instructions to save time
Stop repeating the same instructions to your AI. Cursor's .mdc files let you define your preferences once and for all.
AI can be exhausting: managing frustrations with Cursor
Stubbornness, yes-manning, infinite loops... How to handle moments when AI becomes more of an obstacle than a help.
AI and R&D: where does real human value lie?
AI excels at what it already knows. The innovative part remains human, but the time saved on preparation and evaluation is considerable.
Refactoring code with AI in agent mode: a practical guide
Typing, simplification, readability, paradigm shifts... AI in agent mode is a formidable tool for maintaining and improving your codebase.
Writing tests with AI: the use case nobody wants to do by hand
TDD, end-to-end tests, unit tests... AI excels at writing tests, the work nobody wants to do.
Reading code: the superpower of this decade
AI generates code faster than we can understand it. The real superpower is knowing how to read it and detect biases.
Custom tools and debugging with AI
All those small tasks that weren't worth the effort suddenly become worthwhile. AI also helps with debugging, but not for every case.
Beating the blank page and the myth of building a SaaS in 2 weeks
AI is excellent for starting a project. But beware the prototype mirage: getting it to production is a whole different story.
Data exploration with AI: throwaway code is your friend
Loading scripts, framework exploration, image processing... AI excels at exploration tasks. But be careful to keep only what matters.
Using AI as a framework: lessons learned after 15 years in software
AI behaves like a big framework. It patches together known solutions to solve your needs. And like any framework, you'd rather not have to dig into its internals.
Should you still learn to code? Writing code is good, reading it is better
I'm regularly asked whether it's worth learning to code. I used to say yes. Now, the answer is more nuanced.
AI promises a surge in hacking
AIs have no concept of security by default. Hardcoded API keys, client-side password checks, exposed DB connections... Vibecoding without expertise is dangerous.
Code doesn't matter: eggshell programming
Being a software engineer isn't about writing code, it's about solving problems. Code is at best a hindrance, at worst a source of unnecessary problems.
Jupyter is the poor man's debugger
Inconsistent state, nightmare production integration, zero advanced tooling... Why the Python debugger is the real professional solution.
