Phases
The eight cognitive phases of a unit of AI work
Eight cognitive phases any unit of AI work passes through: Exploration, Scoping, Planning, Execution, Validation, Refinement, Delivery and Reflection.
Overview
How cognitive work decomposes over time
- The framework names eight cognitive phases as the temporal decomposition of any unit of AI work. Exploration opens a piece of work; Scoping bounds it; Planning sequences it; Execution produces it; Validation checks it; Refinement adjusts it; Delivery packages it for whoever consumes it; Reflection extracts what should be carried forward. Together the eight describe how cognitive work decomposes over time.
- This is the temporal complement to the aspects, which describe how a single pattern decomposes structurally. Phases are cyclical and modal rather than strictly sequential. A Cognitive Partnership engagement cycles through them, returning to Exploration when Execution surfaces a new question. A Discrete Delegation task runs them approximately once in approximately fixed order. An Autonomous Agency mandate runs them at two levels: macro at mandate-design time, micro per work item.
- An Encapsulated Utility component runs most of them at design time and compresses Execution and Delivery at runtime. The phases vocabulary lets practitioners describe where the cognitive work is happening at any moment, where in the cycle a particular intervention belongs, and how a pattern's runtime tempo differs from its design-time work. It is canonical framework vocabulary alongside the patterns and aspects.
Contact
Want to talk through the phases of a specific workflow?
Reach out for a conversation about how the phases apply to your cognitive work.
hello@nutropic.ai