Cost Engineering
Seven lessons on deliberate AI workflow spend — routing, caching, per-path expected/p50/p95 costs, bounded loops, calibration, experiments, and a checklist capstone. Account required.
Routing: The Two-Axis Model Choice
Separate the two questions every model choice answers
Caching & Token Budgets
Identify the workflow shapes where caching pays
The Reality Check Loop
Walk the four stages: Predict → Execute → Log → Learn
Path Costs, Branch Evidence, and Bounded Loops
Read expected, p50, and p95 cost for an execution path
Reality Check: Identity, Paths, and Calibration
Match actual runs only to the immutable graph version and hash they executed
Champion/Challenger Cost per Path
Set cost_per_path as an experiment objective
Capstone: The Cost Engineer's Checklist
Apply routing, caching, budgets, and reality checks to one scenario
Measure before you optimize
Upload a usage CSV and get a waste scan of your real spend in 60 seconds. No account needed.
Cost Engineering Certificate of Completion
Complete all 7 lessons (account required) to earn the Certificate of Completion.
Shareable PDF · LinkedIn · Separate from exam-based Cromus credentials
Up next
Operating at Scale