SIMULATED — WORKSHOP DEMO · synthetic borrower · no live RBIH/ULI/AA call · every fact is sourced and labelled
100% · 22/22
Close / the ask

Keep the core. Close the intelligence loop.

A small architecture change creates a large operating question: can the bank measure more evidence-backed decisions without moving the system of record?

Pilot shape
90 days · one state · one bank
Architecture loop

Understand above the core. Record back into the core.

Systems of record
LOS · CBS · LMS · CRM
Intelligence fabric
Identity · evidence · policy · risk
Agent workforce
Research · CAM · EWS · cluster
Human authority
Challenge · approve · account
Decision of record
Sanction · CP · intervention
The core banking system records what happened. The intelligence layer understands what is happening, why, what could happen next, and what to investigate. The human decides. The architecture stays intact.
One experiment, not an MoU

90 days to earn the right to scale

Stop / scale decision

One state · one bank · one product category · one AI credit workforce. Measure a baseline first, run the same work with the workforce, and let the bank’s own committee judge quality.

Days 1–15
Instrument the baseline

Lock definitions, consent path, human authority, current TAT and fully-loaded cost.

Days 16–75
Run the workforce

Shadow then operate on the selected product. Retain every conflict, correction and quality score.

Days 76–90
Committee verdict

Blind-score CAMs, review economics and decide: stop, adjust, or scale.

Pilot scorecard

Measure value without hiding the denominator

Sanction TAT
baseline → pilot

Days from complete evidence pack to sanction

Fully-loaded cost / decision
₹ baseline → pilot

Include officer, review, rework and tooling

Officer throughput
decisions / officer

Count only decisions that pass quality review

CAM quality
blind score

Bank credit committee scores without knowing the author

Conflicts + stale-data catches
count + rate

Caught before sanction, with retraction retained

Scheme capture rate
% eligible cases

Pathways identified and actually claimed

EWS lead time
before policy window

Signal to human intervention, not a default prediction

Rural cluster origination
new-to-credit / anchor

Surfaced per anchor sanction, with evidence path

PSL sub-target movement
percentage-point change

Organic origination, not PSLC purchase

The ask
If it creates no measurable value, we stop. If it does, we scale.