Extracted from ../entities/reference-rate.md.


🟡 Working:

  • Seeding workflow from Benchmark — design the post-migration bulk-load flow for Reference Rates extracted from Benchmark’s Price Book ranges
  • AI confidence thresholds — exact threshold for “high” vs “low” confidence in BR-146; initial recommendations welcome from Oxcon
  • Categorization scope best practices — when should Reference Rates be scoped vs unscoped? Guidance for admin configuration

🟢 Resolved (BR-review, 2026-04-15):

  • Reference Rate versioning — LOCKED. No versioning. Single current rate per Reference Rate is sufficient for v1. Edits overwrite old values.
  • Historical Reference Rate comparison — CLARIFIED. Layer 3 (AI historical comparison) is distinct from Reference Rates. Reference Rates are manual admin-defined benchmarks. Layer 3 AI pulls from Oxcon’s archived Price Books + completed estimates for automatic historical comparison — a different mechanism, not a Reference Rate feature.