At a glance: a scheduled commercial bank running home loan and LAP origination on FINVIJ — AarambhLOS for origination, MoolyaAI for collateral valuation, and NirnayaDecisionAI for decisioning — on one lending ontology, deployed inside the bank's own perimeter.
The problem
Home loan and LAP files moved through the bank the way they do at most institutions: paper-heavy applications, a manual valuation queue running in series with credit appraisal, underwriters assembling bureau reports, bank statements, and GST data by hand, and committee reviews stretching the journey to 7–10 days. Applicant drop-off compounded at every hand-off.
The deeper issue was defensibility. Decisions lived across systems and spreadsheets; reconstructing why a loan was approved — which policy version, which data, whose override — took days of file archaeology. That is a poor answer to give an RBI examiner.
The solution
AarambhLOS — origination on one spine
Digital application capture with eKYC and document OCR across branch and DSA channels, configurable rate cards and business rules owned by the bank's own teams, and maker-checker approval hierarchies with a deviation matrix wired in by default.
MoolyaAI — valuation in parallel, not in series
Geo-fenced, timestamped property capture feeding valuation data directly into the credit file. The collateral track runs alongside the credit track, so valuation stopped blocking the decision.
NirnayaDecisionAI — the decision, with its reasons
Policy rules and scoring orchestrated into approve / refer / decline verdicts with reason codes on every file — clear cases straight through, judgment cases routed to the credit manager with the full data picture attached.
All three run on FINVIJ's shared lending ontology: the borrower the LOS captures is the same borrower the decision engine scores — no reconciliation layer, no re-keying, one audit trail. Deployed on-premise, with zero data egress.
The outcome
Underwriting TAT fell from 7–10 days to 48–72 hours for home loan and LAP files. Every decision now carries reason codes and end-to-end lineage — policy rule to data source to override log — reproducible for any past date. When the next inspection asks "why was this loan approved?", the answer takes minutes, not a week.
The specific numbers above are deployment-level figures shared with the client's knowledge; the institution is unnamed by agreement.
Ask us to walk you through this deployment.
We'll show you the journey, the workbench, and the audit trail — on a live system.
Request Demo