The supply chain finance market has been waiting for the ‘next big thing’ that will transform the product from a means of obtaining early payment on an approved invoice to something far larger that will enable the financing of the entire supply chain.
It is AI technology that will be pivotal to the transformation of the traditional SCF offering, industry experts argue.
“Currently SCF is approved invoice financing. It covers the last 60 days of something that might have started 180 days before with a purchase order. It is perhaps not that impressive,” explains Bart Ras, managing director, UK at Greensill Capital, a company actively exploring opportunities with AI technology and SCF.
“But with this technique, we can get earlier and deeper into the supply chain to support purchase order financing or tier 3 financing as it will provide more visibility and transparency into your ecosystem,” he says.
AI technology has the potential to allow companies and financial institutions to finally tap into the vast pools of data collected by companies’ internal payment systems, such as invoice history and spending patterns. Access to data will enable SCF providers to more effectively assess and price risks, extending more financing to companies active throughout the supply chain.
Machine-learning products could also herald the end for certain repetitive data-entry finance jobs, as algorithms replace human endeavours in supply chain finance – resulting in some people becoming more worried than excited by the AI-powered future.
Yet in practice, using AI in SCF is still in its relatively early stages, with companies currently running test cases, some launching new products or planning a pipeline of solutions to be launched in the coming years.
While the technology has the potential to transform the industry, there are still questions – many of which were raised at this year’s SCF Forum Europe – over how fast and how far-reaching the transformation will be.
The current state of the market
Today, AI is proving popular as a cash forecasting tool, where the technology is enabling treasurers to make more informed decisions based on AI-generated predictions on future cash flows. As these products develop, there will likely be increased crossover with SCF solutions.
In October financial technology firm Taulia launched an AI-powered cash forecasting solution that uses data drawn from purchase orders, payables and receivables to provide insight into future cash flows.
“Rather than just taking the due date – and not everything gets paid on the due date – we can add some intelligence based on patterns we see in that corporate’s data and make a machine-learning prediction on when an invoice will be paid or collected,” explains Vincent Beerman, Taulia’s head of platform data and AI to SCF Briefing.
The AI solution currently available can make predictions on approved payables and unapproved invoices and when they might be paid based on a variety of data points and payment behaviour patterns. The product can also pull in purchase order data and can make predictions on when these orders might turn into invoices.
“How AI differentiates from the standard spreadsheet practices is that AI takes into account behaviour and organisational behaviour. It is dispassionate. That’s the benefit. The model is just detecting patterns,” Beerman adds.
Taulia is still developing the receivables side of this product, which is expected to be added to the cash forecasting tool in the first quarter of next year.
The fintech’s next step is to use AI-powered cash forecasting to help treasurers make more informed decisions about how they use their supply chain finance or dynamic discounting tools.
Next year, Taulia plans to further enhance the power of its platform with a control panel that will enable treasurers to run different AI-powered scenarios to predict either how much cash they could release from their supply chain or how much yield they could gain from discounting invoices. From there they will be able to select which suppliers will be enabled for early payment under which scheme (self-funded dynamic discounting or third party funded SCF).
Fintech Kyriba has been running different use-cases, testing how to use AI in its SCF offerings and other Kyriba products. The company has an initial focus on cash forecasting, with a product due to launch next year offering the ability to predict supplier cash flow shortages.
Based on historical data of supplier financing requests, the system can suggest when the supplier might make future requests, explains Edi Poloniato, global head of working capital solutions at Kyriba.
“It is useful for funders to anticipate credit lines and the amount they need to prepare so they are in a position to fund,” he says.
“The adoption [of AI] will be there. I am convinced it will be a trend for the market to use this as an optional service,” he adds.
Another fintech, the start-up Previse, is already using AI to support supply chain finance. It uses AI-powered algorithms to predict what invoices are more likely to be paid. It separates out those invoices that are most likely not to be paid and enables instant payment to suppliers for all the remaining invoices, minus a fee.
As well as its use as a forecasting tool, AI is also being explored as a means of overcoming some of the regulatory challenges facing the SCF market.
Increasingly, rating agencies and accountancy firms are questioning whether companies are effectively disclosing their use of SCF programmes. There is a question mark over whether SCF should be treated as off-balance sheet or be deemed as debt.
AI could step in here with a solution, explains Ras. “One of the reasons we are focusing on AI is that an Irrevocable Payment Undertaking (IPU) might end up being viewed as bank debt.
“So, it would be great if you could replace the IPU, so that there is no longer an irrevocable payment undertaking by the buyer saying: ‘I will pay no matter what’. Rather we, as a financier, would look at flows and say that we are so confident that this invoice will get paid – we will take the risk. Rather than asking the buyer to confirm it will pay,” he says.
He says Greensill is already starting to do this on a small scale in different programmes.
Has the AI revolution arrived?
So, AI-powered solutions are likely to improve the decision-making process for a treasurer and enable the more strategic deployment of supply chain financing – but will they revolutionize the SCF market?
Doubts or at least questions were raised at the SCF Forum Europe in Amsterdam last month during an AI panel discussion, with audience members asking what happens when algorithms go wrong. Others questioned how fast the switch to AI solutions will take.
Laurent Tabouelle, COO of software provider Codix Group, warned forum delegates not to consider AI as a “magical solution to everything”.
Regulation around the use and collection of data were raised as further obstacles. Corporates looking to work with AI-powered platforms should pay attention to how their tech provider is treating their data, says Kyriba’s Poloniato.
“Ask how secure the platforms are, what the data protection process in the company is and is the company compliant with EU regulations regarding the protection of the data,” he warns.
A common theme among those SCF Briefing talked to is that AI-powered solutions will still require a degree of human involvement. Technology might replace some of the less creative roles in banks and fintechs, but robots are yet to take everyone’s job, they suggest.
AI products will only show you patterns in behaviour, explains Beerman. It is up to the treasurer and the procurement department, which might have more insight into the company’s relationship with its suppliers, to effectively make use of the information generated by the platform.
“Let machine learning do what it is good at – dispassionate analysis – then humans can get in the room and work from the analysis,” Beerman says.
Ras adds: “You should see this like the navigation system in your car. Even if it says turn right and there’s a cliff – you don’t turn right. But it augments your reality with additional information.
“You can enrich your view and add more pixels in the picture,” he adds.
