A business client asking for credit or an account waited three to four weeks while clerks worked through the paperwork by hand, and many gave up before the end. Meanwhile the compliance team drowned in false alarms. We fixed the speed and the noise without a byte leaving the bank.
In a bank this size, onboarding, know-your-customer checks and credit underwriting are a critical bottleneck, and this one was losing business clients to it, as approvals dragged on for weeks behind manual reviews and paperwork.
A business client seeking credit or an account had to submit stacks of legal documents, financial statements and affidavits, and clerks worked through them by hand, a process that ran three to four weeks, long enough that many applicants simply walked away. At the same time, the old risk-monitoring system threw an alert on almost any slightly unusual transaction, so the compliance team burned thousands of hours a month reviewing entirely legitimate activity.
And the client's own information was scattered (across the current-account system, the credit system and the scanned-document archive) so no representative could see a full picture, and clients were asked for the same documents twice.
A secure pipeline, AI that abstracts the documents, and a green lane with a human on anything complex.
We connected the bank's data with external sources (the companies registrar, business-credit data and international sanctions lists) into one integrated platform under strict access permissions and information compartmentalisation.
A language engine reads and analyses scanned financial statements and legal documents in seconds, cross-checks them against the bank's KYC rules, and hands the underwriter a ready risk estimate and decision summary.
Basic credit requests with a high confidence score (above 95%) are approved automatically within minutes, while complex or suspicious cases route to a senior underwriter with a focused alert on exactly the clauses that need a human eye.
The risk layer learns what normal looks like for each client, so ordinary, legitimate transactions stop tripping alarms, and the compliance team gets its hours back for the cases that matter.
Representatives finally see a complete, cross-checked picture of the client, ending the duplicate, redundant document requests that used to frustrate applicants mid-process.
Data is queried under scoped permissions and never copied out, so the whole system meets regulation head-on. The speed came with no exposure of sensitive financial information at all.
Faster for the client, calmer for compliance.
Opening an account and approving credit for business clients fell from three weeks to about 24 hours, lifting the process-completion rate by 30%. The applicants who used to give up now make it to the end.
False positives in the control systems dropped 70%, saving the bank thousands of hours a month and cutting compliance operating costs by about 25%, with airtight regulatory compliance and zero leakage of sensitive data.
The speed came from a simple division of labour: let high-confidence, low-risk cases clear automatically, and spend human attention only where it changes the answer, the same green-lane-and-triage pattern behind our fraud-detection work, where auto-approving the clean cases is exactly what freed people for the suspicious ones. A senior underwriter still signs every complex case, and no sensitive data ever leaves the building.
A European insurer was bleeding money to claims fraud while honest customers waited days. A real-time risk engine cut fraud losses 60% and false alarms 80%, and dropped legitimate approvals from 5 days to 30 seconds.
Read the case → After-call work · FinanceA financial firm's agents burned a third of the day on after-call summaries and CRM entry. AI auto-summaries approved in a glance cut average handle time up to 49% and lifted capacity 35%.
Read the case →
Reef TRH