Knowledge,
not news.
Technical and institutional publications on systems engineering, technological autonomy, auditability, interoperability, portability, and the trade-offs of AI systems. Texts written to remain useful beyond any specific product or implementation.
The Manifest sets out the thesis that guides this work. The Insights that follow develop specific themes, in sequence.
A structured defense of computational sovereignty, digital autonomy, local AI infrastructure, auditable pipelines, and governance of institutional memory.
Publications
Organizations adopting external AI systems transfer more than data — they transfer the accumulated context that makes reasoning about that data possible. Retaining the files is not the same as retaining the intelligence built over them.
Systems not designed to be inspected actively resist inspection. Every layer added without traceability is a layer that will need to be explained by inference, not by evidence — and inference about system behavior is the exact opposite of governance.
An AI answer you cannot verify back to its origin is not an answer — it is a bet. Source citation is not a comfort feature: it is what turns a convincing text generator into a tool whose output can be checked, corrected and defended.
Different models produce different properties. Preserving boundaries between system and model makes it possible to evaluate, compose, and replace components without turning a transient technology choice into a permanent platform requirement.