Artificial Intelligence: What Potential for the Finance Function in Banking?

How can artificial intelligence transform the finance function in banking?
To answer this question, we analysed 9 domains of the finance function, covering nearly 80 processes and 600 tasks, each task drawing on a set of elementary skills to be carried out properly. We then mapped this framework against the ability of artificial intelligence to perform each of these elementary skills.

The findings point to a finance function better equipped to fulfil its core missions.

In short, we identify real opportunities for AI to add value, strengthening the finance function’s analysis and forecasting capabilities and enabling it to operate on shorter cycles and with new depth.

• From processing to analysis →greater command of the numbers

Orchestrating algorithmic and AI tools will eventually make it possible to automate almost all processing, freeing up capacity for investigating exceptions, continuous improvement and analysis.

• Shorter cycles →more responsive steering

Deploying AI-enabled tools shortens production cycles and makes simulations easier to build. In practice, information, whether actual (past) or projected (future, simulations), is available sooner, enabling far better steering and forecasting.

• From delayed to real time →lower risk

Incidents, anomalies and alerts are detected immediately rather than after the fact. Diagnostics can therefore begin much sooner after events occur, and corrective actions can be taken very quickly, limiting the impact.

Opportunities by domain. At a first level of reading, by domain, the potential for AI adoption varies: the largest opportunities lie in Financial Performance Management and Regulatory Reporting, which combine large headcounts with high potential, followed by Accounts Payable and Accounting, which mobilize many FTEs and retain significant potential for AI adoption.

But measuring the impact of AI at the level of individual tasks makes it possible to identify opportunities more precisely, and to single out the processes with the highest potential for AI adoption. The analysis also shows that addressing the top 14 or 15 processes captures 50% of the tasks with high AI potential. These are the processes where the transformation should start, to secure tangible benefits.

Moreover, the opportunities differ in nature but exist across every type of activity in the finance function.

Transactional activities: the opportunities lie mainly in automation and pre-analysis, reducing the data-handling workload and freeing up time for analysis and control.

• Transaction automation
• Stronger controls
• Pre-analysis of exceptions

Areas of expertise: here, the contribution of artificial intelligence will be far more targeted, but opportunities exist in file preparation, scenario development and what we call structured archiving: the ability to fully document decisions and the reasoning behind them, in particular to inform subsequent decisions, whether AI-assisted or not.

• File preparation
• Scenario development
• Structured archiving

Reporting, analysis and forecasting: this is where artificial intelligence can make the greatest contribution within the finance function. Automation, the ability to run many more analysis scenarios at a depth that was not possible until now, and fast iteration and simulation when events affect the business: the potential is significant.

• Reporting automation
• Stronger analysis (depth, accuracy, speed of delivery)
• Multiple forecasting scenarios

Finally, different technologies need to be deployed depending on the nature of the elementary skills involved in each process, as the following diagram shows:

In summary: a best practice is to start with proven use cases, in order to capture business value early and build momentum for the transformation. The table below identifies the “Top 14” processes with the highest potential for AI adoption, the technologies to deploy and the gains that can be expected.