Innovation at the Heart of Treasury: Artificial Intelligence within Everyone's Reach?
​
​
​
Among the elements that make up working capital, there is one that is implicitly more complicated to control than others: Customer receipts. Although all companies should, ideally, be organized to monitor customer invoices and payments, many still do not have the necessary tools for minimum monitoring, and too few are equipped to optimize collection. Excel spreadsheets are still very much in evidence. The teams responsible for collection are frequently not targeted, they use processes that are not very well automated, and which are reactive rather than proactive. In simple terms, the primary objectives of all companies in the area of debt collection are to name customers in arrears and to monitor the history of contacts made with them. However, such practices fail to provide an accurate picture of customer collections.
​
When a company's objective is to improve its cash management and working capital, it becomes imperative to maximize collection efforts and automate as many of the less strategic tasks as possible. This is where Artificial Intelligence (AI) comes in. By analyzing the data available in companies' ERP systems, AI optimizes the efforts of each employee by targeting the customers for whom the cash or risk impact is the most significant for the company. AI also anticipates customer payment problems. This allows the team to take appropriate action before the problems occur. In short, AI helps to automate all the steps that can be automated. This allows teams to focus on tasks that require personalized intervention, allowing proactive actions that alert the team to the risk of non-payment before it occurs. AI will also provide a more accurate view of expected cash receipts, thereby refining cash management.
​
"Ai solutions can easily be within the reach of every treasurer."
​
The greatest concern about implementing AI in a company, whatever its application, is linked to data. Data can be frightening. It is often considered too complex to master, or of too unreliable quality for optimal use. These perceptions, which are probably well-founded, can lead to extreme decisions that oscillate between immobility and a deep cleansing of the data that distracts the business from short-term goals. So, is it necessary to have perfect data to initiate the implementation of AI? Not necessarily! An approach based on successive business "quick wins" is possible. An innovation project for working capital can start by getting the best out of existing data by working through successive iterations. This can involve transforming processes and then enriching the data as the results obtained by the AI are obtained. At each stage, it becomes important to evaluate the cost-benefit ratio between the costs of additional processing of the data and the benefits of a corresponding improvement. In this way, the results appear from the start of the project and begin to optimize the company's working capital in stages. With such an approach, AI solutions can easily be within the reach of every treasurer.