Blog
Thinking
Perspectives on AI implementation, operational intelligence, and the future of work.
AI Agents
Why AI Agents Fail in Production
When agents fail in production, the assumption is that the model failed. In practice, most failures trace to context construction — the engineering problem that decides whether an agent works.
Read the blog →LLMs
Deep Dive: LLMs are an I/O Layer
LLMs excel at parsing ambiguous input and writing fluent prose. They are not calculators, classifiers, or databases. Reliable production systems put a thick deterministic layer between thin LLM I/O boundaries.
Read the blog →Leadership
The First Question a CEO Should Ask About AI
The boardroom question is how to implement AI. The better one is whether the organization is ready — because implementing AI is first a process problem, then a data problem, and only finally a model problem.
Read the blog →Strategy
Where Does AI Actually Fit?
A gate, a map, and four quadrants for deciding where AI belongs across everything you run. The skill is sorting tasks, not businesses — and knowing the one decision a leader cannot delegate.
Read the blog →