SR 11-7 Model Risk Governance
Every agent output is traceable, auditable, and explainable. CAIBots Credit Underwriting AI ships with a pre-built model documentation package, validation protocols, and examiner-ready audit trails designed to support your institution's SR 11-7 / OCC 2011-12 model validation from day one of the engagement. SR 11-7 is supervisory guidance addressed to banking organizations — validation itself is completed and signed by your Model Risk Management function during the pilot.
Platform Design Principles
CAIBots Credit Underwriting Agentic AI is purpose-built for regulated financial institutions. Unlike generic LLM wrappers, the platform operates on a deterministic orchestration layer with bounded agent behaviors, mandatory HITL gates at policy exception thresholds, and full LOS bi-directional integration.
All source tool connections use institution-credentialed API keys vaulted in the customer's own secrets manager — credentials are never persisted in CAIBots systems. Documents are processed in the CAIBots runtime under a signed DPA with 30-day deletion; full data flows are detailed in the Security Architecture Supplement. Synthetic demo data is used in all demo environments.
Agent
Consumer / HELOC
SBA 7(a)
CRE & I
ABL
Construction
Syndicated
DSCR Alert
Renewal
Ag Line
HITL Gate Thresholds by Loan Type
Auto-decision is available only for Consumer/HELOC (score ≥ policy minimum, no exceptions, DSCR N/A) and DSCR Alert monitoring (no HITL required for watchlist flag; HITL required for covenant breach action). All other loan types route to the Credit Officer Workbench for final approval with AI-generated recommendation and full audit trail.