Insights
Long-form articles, guides and field notes on AI engineering, sovereignty, compliance and the human side of security.
Why most production failures of LLM systems are context failures, how tokenization shapes cost and quality, and a practical method for designing context windows as deliberately as you design a database schema.
Sovereignty is decided in architecture, not in contracts: where data physically lives, which law can reach it, who holds the keys, and whether you can leave. A design method for EU-conform AI and platform hosting.
A translation of the Digital Operational Resilience Act from legal text into engineering requirements — ICT risk, incident reporting, resilience testing, third-party risk — with what it means for teams that build and run AI and platform systems.
The classic social-engineering tactics, what generative AI has changed about them, why annual awareness training does not move the numbers, and how behaviour assessments and designed workflows do.