1 How reasoning became visible
This chapter covers
- The hidden cost of implicit reasoning
- The difference between agreement and shared understanding
- Prompts as compact reasoning blueprints
- Five architectures for structuring thought
- The conditions that allow clear reasoning to survive
In many organizations, important decisions appear settled in the meeting room but unravel once work begins. Teams leave believing they are aligned, only to discover later that assumptions, priorities, and trade-offs were interpreted differently. In other cases, the team is genuinely aligned but has converged around the wrong definition of the problem. These failures are rarely caused by incompetence or poor intent. More often, they arise because the reasoning behind the decision was never made visible.
Reasoning has architecture. Conversations can move directly from a question to an answer, expose intermediate logic, branch into alternatives, rotate through different perspectives, or test emerging conclusions against evidence. These patterns can be understood through five reasoning architectures informed by prompting and inference techniques developed around large language models: Direct Q&A, Chain-of-Thought, Tree-of-Thought, Chair-of-Thought, and ReAct.