Taking the Mystery Out of AI in Cash Management

Artificial Intelligence (AI) needs explanation if it is to be more than a buzz word. A recent article in the Asian Cash Management Association (ACMA) journal by Joep Van den Brink from PlanFocus Software threw some light on it in the context of cash management.

AI software uses historic data of seasonal cash activity, monthly payday patterns and identified events in regional and cashpoint group calendars to predict cash demand and the need for cash pick ups and drop offs.

As the data set increases, and changes, the software is able to adapt and learn based on updating its statistical modelling. This is the ‘self-learning’ which is so often mentioned.

The data updates are both from long term data and data that is imported multiple times a day for each cash point, down to the denominational level. If this data is unavailable, or has gaps, then manual input is necessary for event activities.

The algorithms treat cashpoint withdrawal and deposit data separately from any cash recycling and branch device, so that independent flows can be optimally forecasted. In addition to the optimisation of individual cashpoints, numerous specialised algorithms can be utilised to optimise groups of cashpoints such as branches and remote geographical locations, still at the detailed level of each individual denomination and considering intraday regional and group calendar effects.

What AI can deliver

PlanFocus believes AI can:

• Reduce Cash-in-Transit (CIT) and internal replenishment costs

• Reduce cash inventories, particularly important if interest (opportunity) costs are high

• Increase cash recycling, both optimising recycling ATMs and in-branch recycling

• Reduce the flow of cash back to the cash centre, reducing counting and operational costs

• Improve cash agent productivity through fully automating cash planning, ordering and monitoring procedures

• Increase ATM availability (overall ATM and/or cash denominations) resulting in higher customer satisfaction and reduced emergency cash transport movements

• Fully automate and self-learning, removing the need for manual data entry and management.

Today many financial institutions and cash operators rely on relatively simple tools for planning their complex daily cash management operations, primarily Excel spreadsheets. While such tools are cheap to build, they involve manual input, which is labour intensive and prone to input errors, version control and quality control. And documenting how they work can be challenging, as they can suffer from low central control/visibility and easily become obsolete. Perhaps their real cost lies in the fact that organisations may be missing substantial savings.

Why is it important?

Changes in cash requirements mean that optimising cash planning in operations is more important than ever.

In many countries a trend to outsource cash operations to third party providers, or joint ventures, has begun, simultaneously making visibility of data and the management of complexity necessary. To achieve the cost reductions promised by shared service operations across multiple banks requires good data and effective tools to manage it. Studies suggest savings of 15-25% are possible on CIT costs, capital holding/ interests and processing costs.

Finally, one of the prizes of getting it right is a reduction in human intervention, with all the staff costs and risk of errors and losses that brings.