October 29, 2026
The AI Says It Works. The Tests Agree. What Could Go Wrong?
AI-generated code and tests can share the same mistaken assumption. Reliable delivery needs independent requirements and evidence that can expose it.
Journal
Latest Insights & Trends
Read article
More from the journal
October 13, 2026
Your AI Saved 10,000 Hours. So What?
AI productivity is becoming easier to demonstrate. Converting that productivity into business value is the harder problem. Imagine presenting an AI initiative…
Read article →
October 13, 2026
The Dashboard Is Losing Its Monopoly on Data
For decades, we have asked people to navigate dashboards to find answers in data. AI is beginning to reverse that relationship. For…
Read article →
October 5, 2026
Your Data Architecture Is Becoming Your AI Architecture
AI agents need authoritative business meaning, not merely access to more documents. A practical architecture test for building trustworthy enterprise context.
Read article →
September 22, 2026
AI Makes Software Cheaper to Build. What Is Worth Owning?
AI lowers the cost of creating software, but not the cost of owning it. A practical enterprise architecture test for deciding what…
Read article →
September 16, 2026
Stateless MCP Still Needs Durable Business State
MCP’s stateless core makes agent infrastructure easier to scale, but it moves responsibility for authority, progress, and recovery into the application. Here…
Read article →
September 7, 2026
AI Agent Evaluation Must Go Beyond the Trace
Enterprise AI agents should be evaluated against authoritative business state, not merely their execution traces. A practical architecture test for outcomes, authority…
Read article →