Essays and experiments on persistent AI, machine-readable organizations, learning systems, and biological design under real-world constraints.
A same-stream evaluation framework and policy-facing lookup diagnostic for finding memory systems that score well without supporting reliable governance operations.
A proposal for a probationary layer between new information and durable change in persistent AI systems.
Why enterprise agents need runtime controls grounded in business state, not only prompts, permissions, and identity systems.
On memory, consolidation, and the missing decision layer in persistent AI systems.
When assay-explicit biological models earn their complexity by extracting identifiable quantities that improve the next engineering decision.
A failed early-warning study for off-policy fine-tuning under dynamics shift, and the design lessons that emerged from the failure.