Blog

Essays and field reports on reproducible AI, interpretation drift, governing meaning, and the protocol layer. As TCP/AP ships, we'll also write up implementation deep-dives and real-world case studies from the Kernel, the conformance authority, and agents in production.

Essay · May 14, 2026

The frontier AI challenge is not intelligence. It's reproducibility.

As intelligence becomes a commodity, reliability becomes the constraint. Why the next decade of AI belongs to protocol layers and reproducible decisions, not bigger models.

Read →

Essay · May 2, 2026

Substrate Engineering: The Missing Discipline in AI

Civilizations scale on stable substrates, not just powerful inventions. Why the next decade of AI reliability is a substrate problem, governing the environment decisions run in, not just the models.

Read →

Essay · April 21, 2026

Why "Just Code It" Fails in the Age of AI Agents

"If it must be deterministic, why use an LLM at all?" The answer is the interface where messy human language meets governed code, and why the reliable pattern is: AI perceives, code decides.

Read →

Essay · April 9, 2026

Interpretation Drift: The Hidden Failure Mode of AI

Hallucinations get all the attention. The quieter, more dangerous failure is when identical inputs produce different outcomes, none obviously wrong. Why the fix is governing interpretation itself.

Read →