15 Oct 2026
The bottleneck is called middle management ๐๏ธ
Middle managers are not AI enemies: they become one when the system asks them only for coordination and approvals.
Read more โA practical space for people who build software and want to reason better about architecture, automation, quality, and AI. Applicable ideas, structured guides, and a few uncomfortable questions to keep speed from being mistaken for progress.
5 Oct 2026
Generating documentation from a codebase is a skill: coherence, verification, doc-as-code. Wrong docs are worse than no docs.
2 Oct 2026
An AI service subscription does not change work by itself: adoption needs applied learning, protected time, leadership, and a concrete reason to change.
1 Oct 2026
Italy's Law 132/2025 adds no new obligations to AI systems already covered by the AI Act. But it designates two authorities, makes deepfakes a crime and delegates the enforcement framework to the Government. Here is what actually changes.
28 Sep 2026
Greenfield starts from the specification, not the prompt: the spec as the source of truth that outlives the code.
24 Sep 2026
The EU Data Act answers the question almost no maker of connected objects wants to hear: the data from your product belongs to whoever uses it. And it says so in a directly applicable law.
21 Sep 2026
Tokens, output, infinite sessions and rework: the math of what you don't see. When to delegate and when not to.