01Agentic AIWhere should AI agents act autonomously, and where should people stay in control?Start with the consequence of an action, not the capability of the model. A read-only lookup and a change to a customer account need different levels of authority.Read the perspective
02AI deliveryBuilding Production-Ready AI ApplicationsProduction AI applications need approved data, permissions, evaluation, cost visibility and clear recovery paths.Read the perspective
03AI evaluationHow do you measure whether an AI feature is actually working?Accuracy alone hides the operating truth. What matters is whether people complete the task, trust the result and recover when it is wrong.Read the perspective