Partner transaction volumes arrive in bursts — travel-season remittances don't spread themselves evenly. Operations entered them manually, one ticket at a time, which meant queues, overtime, and error-prone re-keying during exactly the peaks when speed mattered most.

Banking constraints, not startup constraints: every transaction must reconcile with Core Banking, RBI reporting is mandatory and time-bound, partner file formats are fixed by the partner (not by us), and a failed row can never silently poison a batch. The ops team lives in Excel — any solution that asked them to learn a new tool would die on arrival.

Met ops where they were: a validated Excel bulk-upload with row-level validation, so a bad row fails loudly and alone instead of blocking the batch. Uploaded transactions flow through the existing Core Banking API integrations (Remittance, LC, BG) rather than a parallel path — one ledger, one reconciliation story.

Around the upload I automated the two adjacent time-sinks: scheduled RBI compliance reports for the Business team and risk-mitigation email alerts that flag potential transaction errors to Operations before they settle.

Transaction turnaround time down 80% on the AD2 partner flow; compliance reporting near-zero manual effort with on-time RBI submissions; proactive error flags instead of post-facto reversals.

Add idempotency keys per row so retried uploads are provably safe; give partners an async status API (accepted / settled / failed with reasons) instead of email round-trips; and ship a partner sandbox with the exact validators production uses, so format errors get caught on their side first.