Executive

Headline levels anchored to published Cagamas results (2025: assets RM56.2bn, PBT RM476.9m, record RM31.3bn raised, TCR ~42.5%). Monthly detail is synthetic.

Share of total assets by product line. Synthetic quarterly series; PWR ~69% and PWOR ~13% reflect the published shape of the book.

Natural language to SQL inside Autonomous AI Database 26ai. Profile EXEC_AI is restricted to the two executive tables (enforce_object_list=true).

Synthetic counterparty labels only (CPTY-nn) — no real institution names. Utilisation above 90% is a review trigger in the demo storyline.

Profile PWR_AI is restricted to the four PWR tables (enforce_object_list=true). Loan-level data is synthetic: no borrower names, IC numbers or real account numbers exist in the schema.

Synthetic model metrics illustrating in-database ML (OML) governance for the PWOR AIRB portfolio — Cagamas applies the AIRB approach to PWOR. CHAMPION = currently promoted scoring model; PSI above 0.2 triggers a drift review.

EL = PD x LGD x EAD over 80,000 synthetic loans scored in-database. Darker cell = higher expected loss.

Synthetic stress scenario — PD multipliers recomputed in-database over the live 80,000-loan book. OPR, house-price and unemployment deltas are illustrative, not forecasts.

Profile PWOR_AI is restricted to the four PWOR tables (enforce_object_list=true). Loan-level data is synthetic: no borrower names, IC numbers or real account numbers exist in the schema.