Artificial intelligence faced with cash forecasts

TAIGA responds to a journalist from Les Echos

A journalist from Les Echos [1] recently questioned the question of what could be useful cash forecasts, while appearing to want to confront the cash and artificial intelligence.

It seems that “the point of view of most of the chanters of new technologies” converge to that of corporate treasurers to be satisfied with a cash forecast that would simply consist of accounting entries that have not yet had their Therefore, the author is quite clear that the treasury software can indeed be satisfied, by primary purpose more than by a biased ideology of the constitution of a forecast cash, d ‘such a result, that the corporate treasurer is then quite legitimate to exploit in order to cover its possible needs at the shortest term.

This very short-term vision of the forecast, which could in other words be called a reality already foreseen, seems to be at this stage the only (or first?) Horizon of our journalist, who concludes that before effectively treating the subject of the forecasts, it is necessary beforehand to answer the questions related to the objectives, the stakes and the role of the treasurer in the constitution of the methodology which will allow with or without artificial intelligence (here, the subject is covered …) to envisage cash flow . We all share the opinion of Pierre Dac, who reminded us that “forecasts are difficult, especially when they relate to the future”.

“Forecasts are difficult, especially when they are about the future”.

If the first role of the treasurer is obviously to manage the cash flow, the latter in its daily reality can not ignore the flows already noted, nor on the soon-to-be-observed flows. This is his reality today, tomorrow morning but also sometimes retroactive reality, its banking partners do not deprive themselves of the exercise of retroactive flow!

However, limiting the satisfaction of this great money-maker of the company to the vision of proven or almost proven flows seems reductive. If to govern, it is to foresee, to anticipate, it is already to govern, and the treasurer has ceaselessly to know what will happen, in more or less long term. This makes it possible to respond to strategic issues, it also allows and sometimes especially to exist.

Fortunately, in this way, the fintech or new technology players – membership of a group is not important – have been looking at the cash flow forecast for some time now, and this is an opportunity to put into perspective innovation in this area, and to provide some answers to the questions asked.

The subject is far from being just a question of the flow of Data and their processing

A first articulation of response would be to define the role of the treasurer compared to that of management control in the production of cash forecasts. The stakes are not the same, and the vocabulary used in this area either. In fact, if the former is more interested in the cash and disbursement flows themselves, it is because he is primarily responsible for the proper use of the funds available, in accordance with the authorizations negotiated with his partners. financial, banking or otherwise. He anticipates to cover his positions, and wonders about the movements of tomorrow, often pushed to a quarterly horizon to have time to act.

Depending on its context, it will be more or less at heart to refine the details of its forecasts to cover a need for a month, a week, or even a day. Negotiate, alert, cover, anticipate and react. These are some of the objectives identified, it is up to him to organize himself to anticipate.

A second element of answer is brought on this notion of organization. Is the treasurer alone at the head of a company, or does he manage his treasury through subsidiaries, in France or elsewhere? According to the answer, the organization of his work will differ, to rely in the last case on a team. It is then that a clarification in the definition of flows becomes important, so that everyone works harmoniously, with the same language and the same definition of these cash flows. This allows our treasurer to consolidate elements of the same nature, and to avoid any confusion between accounting flows or flows of a financial nature.

A third element of answer is carried by the methodology applied in the very constitution of the forecast, which the treasurer thus privileges in the short term. His more natural approach to flows necessarily leads him to put his forecasts against his achievements. In a pragmatic way, he will tend to rely on his knowledge of the flows, their seasonality in a week or a month to try to reproduce the future movements and their cycles of realization. By confronting the reality with the forecast, he will learn from it, in a very pragmatic way, to help him improve the quality of his forecasts.

If the treasurer has as its benchmark and benchmark its cash balance, the controller management, it is rather the reflex to look for terminals in the balance sheet of his business, with the difference that can exist between one balance and the other. Because the management controller, he, at a different level, will develop all his art to forecast a budget, sales and purchases.

One controls its risks by controlling the cash flow, the other by controlling its operating profit. One can see the wall arrive, the other no, because, unfortunately, the translation of a cash sale is sometimes neither immediate nor obvious.

However, each in its place, the management controller will build a budget and produce all the elements that will allow the management of the company to make the right decisions. Anticipate stable resources, to which the treasurer will necessarily be associated, to make good use of them, and give a cash translation to his budgets, to ensure that the budget is realistic and balanced. And this is where the main question of our subject, which is to reconcile the treasury budget of management control with that of the treasurer, arises. It is this relay which is fundamental and for which the artificial intelligence will never bring anything. Our subject is a question of vocabulary, definition, methodology and also of will and organization.

The world of fintechs and new technologies at the service of treasury is booming. New entrants come, each in their own way, to contribute innovative elements of response to cash flow issues (cash management, cash flow forecasting, payment methods, fraud control, financing) without, for the moment, the convergence is perfect, if not already true complementarities. No one will come to converge the different actors of the same company, all in the service of the forecast, otherwise than with the method and the will.

The subject is far from being just a question of the flow of Data and the processing of it, and the risk is very great to remain in the vague and confused, with or without the help of Artificial Intelligence, without doing evidence of a serious methodology. [i]

On the other hand, in terms of innovation, Artificial Intelligence must improve the relevance of the forecast, and it can only be done through the exploitation of very large data flows.

Management of uncertainty

What is missing from these devices is the management of uncertainty. A forecast is essentially an estimate and is therefore not 100% reliable. Uncertainty depends on the punctuality or otherwise of suppliers in their payments, etc. By supporting this device on Artificial Intelligence (AI), the uncertainty would decrease: a historical analysis would make it possible to estimate the probability for a particular supplier to pay late or not, or for the bank to process the receipt in good order. date, and he would seek the right information, in quantity and quality, in the commercial database. Moreover, this new device would reinforce its initial setting by refining it through continuous experience (the principle of “machine learning”). The IA would therefore provide a guarantee of completeness,
In addition, the addition of AI does not induce a revolution in the information system and only requires read access to different databases. The parameterized algorithms and the indicators that it will build will be used to make the cash flow forecasts more reliable. Management of the liquidity of the company or group will only be improved. [Ii]

Integrating exogenous elements into the modeling of sales or purchases, collections and disbursements, this is how Artificial intelligence will support the superior intelligence of man.

[1] Artificial intelligence and cash: forecasts for what? Les Echos November 12, 2017 by Stéphane Robert

[i] Learn more at


Etienne Téqui

1. Artificial intelligence and cash: forecasts for what? Les Echos November 12, 2017 by Stéphane Robert

Articles similaires