The document-heavy, rule-heavy work that Indian practices spend most of their hours on.
Splits a large transaction population into sub-populations that share characteristics — value band, approver, location, posting route — so the auditor can size and draw a sample within each, and separately surfaces items whose attributes argue for specific testing outside the sample altogether.
Pulls merchant, date, total and tax off a photographed or scanned receipt, fills the claim line, and holds back the ones that fail a policy test — a duplicate image, a missing supplier GSTIN where credit is being claimed, a date outside the approved travel window — for a human approver to settle before the payment run.
Scores general-ledger postings against the attribute patterns that make an entry unusual — rarely paired accounts, a poster who does not normally post, a date after period close, a round value, no narration — and hands the auditor or the controller a ranked queue whose underlying support is then pulled and inspected by hand.
Matches every purchase invoice in the books to its line in the monthly GSTR-2B statement, groups the breaks by cause — supplier has not filed, wrong GSTIN, wrong tax period, value mismatch — and drafts the follow-up each supplier needs so the break is fixed where it was created rather than argued about at assessment.