← ALL CASE STUDIESAI & DATA PLATFORMS · 2026

Bank onboarding, automated.

How a retail bank cut account opening from six days to eleven minutes with document intelligence the auditors actually signed off on.

CLIENTRetail bank
SERVICESAI & data, custom software
TIMELINE9 months, team of 5
HEADLINE RESULT6 days → 11 minutes
RJP-2026-003 · ONBOARDING PIPELINEALL SYSTEMS NOMINAL
11 MINACCOUNT OPENING · MEDIAN · WAS 6 DAYS
INTAKE ▸ OKDOC-INTEL ▸ OKRISK SCORE ▸ OKREVIEW ▸ OK

01The challenge

Opening an account meant six days of manual document review across three back-office teams. Every applicant document — IDs, proof of address, income statements — was read by hand, and a third of applications stalled on avoidable data-entry errors. Applicants abandoned; competitors onboarded them the same afternoon.

02The approach

We built the unglamorous 90%: an intake pipeline with document classification and extraction, a risk-scoring service with human review only where the model is uncertain, and full audit trails on every automated decision. The compliance team helped design the evaluation suite before a single model went live.

BEFORE
MANUAL REVIEW
3 TEAMS · 6 DAYS
AFTER — AUTOMATED PIPELINE
INTAKE
DOC-INTEL
RISK SCORE
REVIEW

03The result

11 minMEDIAN ACCOUNT OPENING
97%STRAIGHT-THROUGH PROCESSING
0AUDIT FINDINGS TO DATE

The bank's onboarding team moved from data entry to exception handling. The evaluation suite runs on every model change, and the audit trail satisfied the regulator on the first pass.

"The first AI project our risk committee approved without a fight — because the guardrails were designed before the model."
HO
HEAD OF OPERATIONS, CLIENT NAME
NEXT CASE STUDYGrid telemetry at scale
Start a project