The workflow, plainly
A member deposits a foreign-currency check drawn on a non-Canadian bank. The branch captures the item, but from that moment it enters a manual corridor: the item must be physically handled or imaged for a correspondent, the exchange rate must be locked at the right moment, GL entries must balance across a suspense account, and Reg CC hold notices must go out on time.
In most institutions we have studied, this workflow touches three distinct systems (core, GL platform, and a correspondent portal), crosses at least two compliance checkpoints (BSA screening and Reg CC holds), and contains a manual reconciliation step that is assigned to no single role.
Why it gets stuck
The failure pattern is consistent: the check arrives at the branch, gets logged, and then enters a queue where it waits for someone with correspondent-portal access. That person may not be in the branch. The exchange-rate lock happens at a different point in the process than the GL posting, which means the amount that posts to the member’s account and the amount that settles with the correspondent can diverge if the rate moves.
Reg CC hold notices have specific timing requirements. When the manual steps stretch to day three or four, the hold-notice window compresses and the compliance risk rises. The reconciliation step — matching the GL suspense entry to the correspondent settlement — is often done weekly in batch, which means discrepancies surface days after the deposit.
What regulation it touches
Regulation CC (funds availability and hold notices), BSA/AML screening obligations for cross-border items, and general GL accuracy requirements under your institution’s own policies. We stay at framework level here — verify your specific hold schedules and screening thresholds locally.
What good looks like
The target state is not “fully automated.” It is a workflow where: (1) the exchange-rate lock and the GL posting happen in the same transaction, not hours apart; (2) the Reg CC hold notice generates automatically from the deposit event, not from a manual trigger; (3) the correspondent submission does not require a person with specialized portal access to be physically available; and (4) the reconciliation runs daily, not weekly, with exceptions surfaced same-day.
This is a bounded, well-defined problem — exactly the kind of workflow where a Prove It Sprint can quantify the leak and define the build boundary before anyone writes code.
