The methodology that drives the AI.
Generic AI guesses at what to sample. Klerum applies your methodology. Your sampling approach becomes a reusable, editable engagement profile that the AI follows the same way across every bank, every engagement.
Firm default, institution override.
Define your methodology once at the firm level. When a bank needs something different, an institution-level profile overrides the firm default, and the AI samples accordingly.
- Coverage targets, such as 30 to 35 percent of outstanding balances
- Exposure-tier selection across the commercial portfolio
- Classified (RR 7) and criticized (RR 6) handling with configurable codes
- Weighting toggles: Watch credits, new or renewed loans, fully-drawn lines, participations
- Loan-officer spread above a minimum portfolio threshold
Every bank rates its own way.
Seed an institution with the standard 8-grade template or a fully custom schema, whether 1 to 10 numeric, OAEM, or whatever a given bank uses. The AI and the reviewer both work in that bank's own rating language everywhere its loans appear.
Chosen at sample time, persisted on the engagement.
When a reviewer runs an AI sample, they pick which profile applies for that run. The choice persists on the engagement, so every reviewer who touches it works to the same standard and the AI never drifts.
See Klerum on your own portfolio.
A 30-minute demo, no deck. Upload a sample trial and watch the AI propose a review population against your methodology.