Cash is one of the largest physical costs a bank carries — idle balances, CIT contracts, insurance, branch effort — and it usually lives in twenty spreadsheets that never agree. AMI puts branches, vaults, CIT and central bank movements on one spine, so the cash position is a number, not a reconciliation project.
The first question is the one most banks can't answer before lunchtime: how much cash do we hold, where is it, and in what denominations? One command view across branches, vaults and the network — refreshed from the operation, not typed into a template.

Every cash view — branches, vault, CIT, forecast, regulator — reached from a single hub.
Branches hold cash the way they always have — by habit. Fleet-style health scoring shows which branches sit on too much, which flirt with running dry, and which are drifting from their own baseline before it becomes an incident.

Branches approaching their holding limits, ranked by urgency — so remittances get scheduled by risk, not by rota.
Idle cash earns nothing, CIT trips cost money whether they're needed or not, and nobody's budget line says "cash we didn't need to hold". Pricing the whole operation — carry, movement, insurance, effort — turns cash from a habit into a managed cost.

Carry cost, CIT, insurance and idle balances — what holding cash costs, per branch and for the network.
Yesterday's position is bookkeeping; tomorrow's is money. Demand forecasting per site — the same discipline AMI applies to ATMs — means replenishment is planned against predicted need, and deviations from forecast become the alarm, not the surprise.
Cash doesn't stop at the branch door. On one side sits the central bank — draws, returns and the reporting that must reconcile. On the other sit corporate clients who deposit and draw cash as a service. Both edges live on the same spine as the branches.

Draws and returns against the central bank tracked continuously — the report agrees with the operation because it comes from it.
These screens run on real operational data. The fastest way to judge the fit is to see the same dashboards built on a sample of yours.