The irreducible epistemic atoms underlying the curriculum. 4,828 atoms across 8 types and 2 molecules
Run independent agents (those with no input dependencies on each other) in parallel to compress total execution time, and serialize only those agents where one genuinely requires the other's output.
Distinguish artificial serialization (treating independent tasks as dependent) from premature parallelization (launching tasks before their dependencies resolve) by mapping actual information flow rather than assumed ordering.
For each pair of adjacent workflow steps, ask 'Does step B require the completed output of step A?' - if no, the steps can run in parallel; if yes, they must be sequential.
Before optimizing any workflow step, measure actual elapsed time across three to four execution cycles rather than relying on felt difficulty or estimated duration, because subjective experience systematically misidentifies bottlenecks.