Blog
Thoughts on building agentic AI systems that work in production.
The AI Value Gap Is a Traceability Problem
Eighty percent of people using AI say it made them more productive. Six percent of their organizations see significant earnings impact. What separates them is not model capability — three-quarters of the high performers redesigned the work itself
Read article →What Two Failed Deployments Taught Me About Governing AI Output
I shipped the same defect twice in two days and a person caught it both times. What fixed it was not better prompting or more careful review — it was a deterministic check that refuses the deploy. A field note on measuring the artefact, not
Read article →Building a 46-Service Autonomous Ecosystem with Claude Code
How a defense systems engineer built a 48-service autonomous AI ecosystem using Claude Code, and what it reveals about agent reliability, governance, and operational discipline.
Read article →More articles on multi-agent architecture, AI governance, hybrid LLM routing, and production AI discipline coming soon.