Poll Results: Use of AI in Cash Forecasting

In April 2024, CXC asked 141* corporate treasurers about their current and planned usage of AI (Artificial Intelligence) in their company’s cash forecasting process. Here are the results.

Ten percent of respondents said they were using AI in their cash forecasting and a further 17% have plans to do so. Sixty seven percent are interested in AI for forecasting but have no plans in place and seven percent are not interested.

A report based on the approaches and experiences of treasurers using and planning to use AI in their cash forecasting will be published in the next few weeks. Please register your interest in this report here: 

A number of comments were also received which are listed below.

Using AI in Cash Forecasting

  • We have built up things to forecast collections through AI, but not fully for cash forecasting
  • pilot stage
  • Started in a few regions as part of a global rollout
  • It’s a mix between 1 – 2 I guess, we have pioneered with AI in our cash forecasting process but could not extend this because of some resource issues. Planning to pick this up when resource constraints have been sorted

Planning to use AI in Cash Forecasting

  • 2 (but will take time)
  • It might also be interesting to know if use or planned use is built or bought - i.e., as part of an internal AI/data analytics tool or integrated as part of a TMS/other tool.
  • So far we found of cash flow uses cases, and have identified that the actual AI code is likely extremely simple (most likely 5-10 lines of code). The actual work seems to be in the data, cleaning, making it meaningful (AI on un-meaningful data = un-meaningful result).
  • Comment: we leverage an internal product for cash flow forecasting, which we plan to upgrade with AI functionality in the future

Interested in using AI for cash forecasting but no action to date

  • We have communicated with our banking partners on their capabilities, and will be demo’ing a solution by a large bank in the upcoming weeks. It just takes time to set-up rules and monitor, and we have been short staffed, but we do want to evaluate the solution to see if it presents any benefits to the cash forecasting process
  • I'm definitely interested in this topic. We run a highly decentralised Group, so compiling a consolidated cashflow forecast would be exactly that; we'd need each individual entity to compile a cashflow forecast and then we'd need to consolidate them. The benefits of doing this would not outweigh the amount of time and effort collectively spent on it, hence we don't do a cashflow forecast. The only way we could ever make it a worthwhile exercise would be to have an automated cashflow forecast run off standard data (like AP/AR reports) and then learning based off actuals. This would obviously have some cost in software but wouldn't involve as many man-hours. It's hard to make the case to do this though, and certainly to find the time to do it!
  • 3 for now, however, hopefully a 2 within the next 12 months
  • But I use AI (CHATGPT) extensively if I have to write longer emails on certain subjects, for example what steps to take before you can issue a dividend.
  • I haven’t seen anything groundbreaking yet but am interested in seeing what it can do
  • We are 3). CF forecast is done only when a funding is required
  • Would be interesting to understand how use of AI versus machine learning tools is distinguished
  • Between numbers 2 and 3 but closer to 3.
  • As an AI based SAAS company we would like to give it a try for such an AI based cash management tools (ex: Tesorio & Atlar).
  • but noting unlikely to take any action in the medium term (3 years?)
  • We are in category (3) and will include this in our S4H journey

Not interested / relevant

  • The AI forecasting tools I have seen so far rely on historical data and trends to determine future cash-flows. Whilst historical data and trends might be of some help to fix baselines there are too many differences between annual and monthly cycles to gain benefit from AI for cash forecasting

*This included 24 who responded via LinkedIn - of these, the results from 17 being corporate practitioners were included, votes from the remaining 7 comprising consultants and bankers were excluded from these results

 

 

Date posted: 
Tuesday, 23 April 2024