Signal & State is a weekly technical publication for technical leaders building AI systems that survive production. It focuses on state, retrieval, orchestration, evals, observability, recovery, and human control.
Most AI products do not fail because the first answer looked bad. They fail later, when the surrounding system has no reliable state, no bounded execution model, no observability, no recovery path, or no clear way for humans to steer what is happening.
Signal & State exists to make those production seams explicit before technical leaders turn a promising prototype into product infrastructure.
Signal & State is written and edited by Ferre Mekelenkamp, a hands-on technical engineer with 10+ years building and shipping production software, including Laravel applications, AI-enabled products, and internal systems.
The publication is written from the builder’s side of AI adoption: the code paths, approval gates, incident surfaces, data boundaries, and team workflows that decide whether an AI feature becomes dependable software.
The standard is simple: make authorship, purpose, method, and contact paths visible without turning the page into marketing sludge.
Issues begin with a production question: what state is trusted, what action is allowed, what failure is recoverable, and what humans can inspect.
Source-backed claims, implementation details, and architecture judgment are kept distinct so readers can tell what is observed, inferred, or recommended.
The publication does not chase word counts, trend recaps, or “top tools” lists. The goal is useful engineering context for technical leaders and senior builders.
Corrections, missing context, and reader challenges can be sent directly to the author by email.
Architecture breakdowns for CTOs, technical founders, and engineering leaders building AI systems past the demo. No hype, no model-release recaps, no “top 10 AI tools.” Unsubscribe in one click.