Banking » Fintech » How finance can use AI strategically

AI is often broken down into three families of applications: (1) machine learning, (2) natural language processing and (3) robotics. They all derive their “intelligence” from advanced statistics technics.

What AI can do intelligently as compared to a human being is extremely limited. It does not have a conscience, it cannot understand a text to summarise it, it cannot hold a conversation (the Turing test). As such, AI is not really “intelligent”. But it can certainly deliver huge support to humans thanks to the ability of computers to store vast amounts of data and compute faster and more accurately than humans.

So, the combination of humans and machines is what matters.

AI can impact your business strategy by helping the creation of more intelligent organisations. It will not develop autonomously, on its own… It is the humans in the organisation who will create the improved interactions between humans and machines.

The right people combined with the right computers and systems will create the most “intelligent organisation”. To prove this simple point, a human having access to Google Search and Wikipedia is much more intelligent and capable than on his/her own.

Applications of AI vary enormously in scope and nature and it is down to every company to work out what is best for them.

It is worth narrowing down the field of reflection, and provide clues to CFOs and FDs on immediately applicable solutions based on AI to (1) improve revenue and (2) reduce costs.

Improving revenue

The revenue line of a P&L is the aggregation of all the individual transactions generated by the invoicing department. Each individual transaction carries lots of information on the customer, the product or service sold, the price, the timing of the transactions and its payment, even the salesperson’s  name, the supply chain involved, etc.

This vast amount of data is what AI loves to absorb and exploit. By structuring the data in a proper way, AI can help find patterns of behaviours and data structures to answer questions around issues such as price sensitivity and elasticity, profile of the most profitable customers, profile of the most profitable product or service, the most profitable geography, the best sales man, etc, Trends of sales pattern can generate description of behaviour, and prediction of future trends or tastes. From there, the company can be more proactive in developing new products or services.

The data can be cross referenced to external information like weather or economic conditions (inflation, FX, etc), competitors offering, and much more.

The cash collection process can be supported by an AI tool analysing every customer profile and supporting the rebate applied to each customer, the likely behaviour for payment, the whole risk pattern.

A customer service department answering calls from potential customers can also benefit from automated chat bots. These are advanced speech recognition systems who try to interpret the questions raised by a customer in parallel to the customer service employee. The chat bot can then suggest answers to the employee to accelerate the answering process. At this stage, a chatbot cannot be on its own, there must be a human to interface with the customer.

Reducing costs

The range of possibility is endless, but I wish to list a few here:

  • Procure to pay:
    • Image recognition systems to recognise supplier Invoices and key-in all the necessary information into the account payable system in minutes. It can process hundreds of invoices per hour, saving that much manual work.
  • Record to report
    • Smart data analysis to reconcile bank accounts automatically
    • Support to report in multiple GAAPs: IFRS, US GAAP etc
  • Order to cash:
    • Smart data analysis to allocate cash received to the right invoice
    • Analyse customers’ drop out and recommend process improvement
    • Automate the “decision tree” leading to the optimal cash collection strategy customised per individual customers
    • Chat bots for after sales queries
  • FP&A:
    • Produce assisted budgets and forecasts relying on smart prediction algorithms
    • Associate KPI and dashboards with natural language processing to prepare performance analysis in words
  • Audit:
    • Data analytics applied to bank accounts, accounts Payable, Payroll, etc… to identify double payments, abnormal amounts, fraud…

All these applications are immediately available. They are part of trusted, validated offerings from several software companies. The path to these achievements is to take the time to assess all different departments of the company and have an AI awareness to identify existing manual and repetitive work and replace them with smart solutions.

The use of AI strategically is a mindset needed at the executive team level. These technics are not necessarily expensive. Some of these projects will have a return on investment within one year.

In our competitive environment, the best performing companies will be those who understand and exploit AI to their immediate benefit. Those who do not will lag.

I wish you all the best in your AI journey.

Guillaume de Pommereau was CFO of Hitachi Europe from 2014 to March 2020. During his tenure, he implemented several of the technologies described in this article with significant improvement in the efficiency of the team. He studied AI and its business implications at MIT Sloan School of Business.

Download our Whitepapers
Accounting Software
Finance Process
Finance Process
FD Surveys