Credit Approval
with AI Analysis

The group processes over 1,000 credit applications per year across all divisions. Each application follows a six-level approval chain — Sales PIC initiates, the credit executive prepares financial analysis, Sales Manager and Sales GM add recommendations, the credit head provides analysis, and the MD gives final approval.

This entire process ran on email and physical documents. Supporting documents — audited financial reports, credit bureau reports, trade references — were attached to emails or physically printed and routed between offices. Email attachment limits capped at 25MB, making multi-document submissions fragile and incomplete.

Physical sign-off at senior levels meant applications sat in transit, untracked. No one could see where a specific application stood without a phone call. Approvals took weeks. The most time-consuming step — the credit executive manually reading dense audited financial statements and computing financial ratios — sat at the centre of every single application, over 1,000 times a year.

The solution was delivered in two distinct phases, each addressing a different layer of the problem.

Phase 1 digitised the process — replacing email initiation, physical routing, and manual tracking with a governed low-code platform workflow covering all six approval levels. The goal was to make the process reliable, visible, and auditable before any intelligence was applied to it.

Phase 2 layered AI onto the digitised foundation — using a local language model to automate the most time-consuming analytical step: extracting financial figures from audited reports and computing the standard ratio set. The credit executive shifts from manual computation to review and verification. The AI analyses; the human decides.