Read this first: everything below is hypothetical and synthetic — the company, the lenders, the numbers and the timeline are constructed for illustration. AssureLens is at design-partner stage; this scenario mirrors what its engine computes on the synthetic demo borrower, not a customer deployment. It is published as a thinking aid, not as evidence of outcomes.
The setup
“Sri Velan Auto Components” — our synthetic borrower — is a Chennai auto-component maker with ₹22 crore turnover. Its banking relationships, seen one at a time, are unremarkable: a ₹50 lakh term loan and a ₹30 lakh working-capital line with Lender A (both regular), a vehicle loan reported on the bureau, and a fresh ₹20 lakh facility under a bank–NBFC co-lending programme. Each lender's book shows a performing account. Each periodic review, run on its own sliver, reads clean.
What the structural map shows
Assemble the same borrower once, with every figure labelled by source, and the picture sharpens without a single new data right:
- Exposure by class: ₹63 lakh lender-known, ₹18 lakh programme, ₹7 lakh bureau-reported — presented side by side, because the classes may overlap and no honest total exists. The point is not one number; it is that three sources describe three different slices.
- A potential collateral overlap: the same Chennai property reference appears behind the term loan and behind a bureau-reported facility — different facility sets, medium confidence. Perhaps a refinance never released; perhaps a data error. Either way it is now a named question with evidence attached, where yesterday it was invisible.
- Named gaps:no consented bank-statement data, no MCA filing check in scope. The review knows what it does not know — which is precisely what a “clean” single-lender file conceals.
What velocity adds
Ninety days before any payment is missed, the rate of change turns over: two facilities opened across two different programmes inside a month, plus a refinance — 300% of this borrower's own baseline. Debt is up 47% year-on-year against revenue up 10%; capacity-adjusted growth fails the configured margin. Mean repayment delay has crept from 1.3 days to 14.5 — no bounce yet, nothing a bureau cycle would flag as delinquent.
Individually, each of these is dismissible. Together, carried by one explained alert — metric, window, baseline, sources, confidence, limitations — they are a review trigger with a specific agenda, weeks or months before the first cheque returns.
The honest counterfactual
Would every lender have missed this? Not necessarily — a sharp analyst with time and the right instinct pulls the same thread. The claim is narrower and more defensible: the assembly that takes that analyst a day of collection happens continuously, the comparison against the borrower's own baseline happens arithmetically, and the result arrives explained and auditable. It is designed to reduce review effort and surface accumulation earlier — a pilot target to be measured against a lender's baseline process, not a guarantee. And the decision, at every point, stays with the lender.
Where a real version of this begins
A shadow pilot: the lender's historical book replayed through the engine, zero integration, results compared against what actually happened. That comparison — how many run-ups would have surfaced early, at what alert volume — is the honest test of the whole idea, and the first thing we run with a design partner.