ES
EN
DE

Articles

Governance and risk

What I learned analysing four security failures that looked unrelated

Four independent incidents are four patches. Four incidents with a common cause are one missing control.
Business and strategy

Estimating an AI project when half the scope is a question

Budgeting an experiment as if it were a task is the most common reason an AI deadline breaks.
Coursework

The feature that was not in the rubric

I added an audit log to a coursework project nobody was going to audit
Evaluation and method

I wrote the limitations of my analysis on the same screen as the charts

Declaring the limits of an analysis is not false modesty. It is the part that makes the rest usable.
AI engineering

AI agents in production: the question almost nobody asks

An agent that fails completely gets noticed. One that fails halfway leaves the system in a state nobody designed.
Evaluation and method

How to tell whether your decision matrix holds up or falls over at the first push

The honesty of an analysis is not in the number. It is in the interval.
Governance and risk

Switzerland has no AI law — and that sentence is where due diligence starts going wrong

The EU AI Act applies by market, not by registered office. A Swiss startup can be inside it without knowing.
Evaluation and method

Why optimising for MAPE can cost you money without you noticing

Choosing the loss function is the analysis, not an implementation detail.
Coursework

Why proving something is better is not what makes it happen

The thesis proved the environmental case. That was the least useful thing it found