Rimilia

Company Background:

Rimilia Holdings Ltd (Rimilia) design, develop and implement intelligent finance solutions, that interfaces simultaneously with Banks and businesses’ Enterprise Resource Planning (ERP) systems to provide intelligent automated processes. The company provides 4 types of solutions to cater for all aspects of Accounts Receivable Automation. ‘Rimilia Cash’ being at the heart of Rimilia products, is a unique solution in the market that intelligently automates the otherwise time consuming, cumbersome and error-prone process of invoice matching and cash allocation that deliver between 60-80% cost-savings to clients.

More info can be found on their website at www.rimilia.com

The Challenge:

As part of the cloud service, Rimilia was processing a great worth of invoicing, payment and credit control data across their clients’ customer base, which showed their debtor’s payment history.

Rimilia had previously used this data to develop software that attempted to predict payment behaviours for a handful of debtors, for example, whether a customer pays on time, or waits for some form of communication to prompt payment. However, there wasn’t enough in-house expertise to accurately evaluate the effectiveness of this software and potentially replace (or improve) it with the state-of-the-art available technologies to assess and model their clients’ customer base.

The Solution:

Aston is well known for its cutting edge research in the fields of artificial intelligence, computer and data science, as well as world class academics and researchers. The challenges faced by the company were addressed by the means of Aston and Rimilia’s collaborative project – Knowledge Transfer Partnership – a Government funded initiative that allows companies to access the expertise of a University’s academics.

This particular KTP project aimed to develop capacity within Rimilia to:

Predict the invoice payment dates for its customers based on historical data of their behaviour.
Identify well and poor behaving debtors to help with defining collection or customer relation strategies.
Predict the payment patterns of the debtors of its customers at an aggregate level.
Monitoring debtors’ predictability and fluctuations to identify risks and opportunities.

The academic input during the project enabled the company to identify patterns and clusters in their datasets, to predict a much higher percentage of debtor behaviour, evaluate those behaviour and subsequently inform the development of the new software module ‘Rimilia Insight’. Linking the back office Accounts Receivable (AR) function to Sales on the front end with real time insights on customer payment ensuring businesses maximise their revenue opportunity is a key underpinning to Rimilia’s drive around Financial Relationship Management (FRM). In addition, predictability of cash flow to the Treasury function allows our clients AR teams to truly become a key internal business partner as well as a centre of excellence.

The Rimilia/Aston Team:

The KTP team worked closely with the data analysts and software development team at Rimilia. The emphasis of this project was to ensure that knowledge was shared between the Aston academic staff and the in-house Rimilia personnel.

The Aston KTP team was led by Dr Maria Chli, who is a computer science lecturer within the School of Engineering and Applied Science and director of Aston’s Computer Science Industry Club and Dr Yordan Raykov, Lecturer in Mathematics.

Dr Chli’s research interests lie in the areas of artificial intelligence. She is particularly interested in applying machine-learning techniques to intelligent agents and multiagent systems, with a focus on smart cities applications.

Dr Raykov’s is a member of Aston’s Non-linearity and Complexity Research Group. The Non-linearity and Complexity Research Group has high international visibility in the areas of pattern analysis, probabilistic methods, non-linear dynamics and the application of methods from statistical physics to the analysis of complex systems.

The project was run by lead Data Scientist and the KTP Associate Dr. Ali Rezaei Yazdi who spent a part of his time at university liaising and consulting with the senior academic staff, named above, and part of his time at Rimilia researching and developing the solution.

Ali has since joined Rimilia full time after completion of the KTP to further enhance Rimilia’s AI capabilities and roadmap. He’ll work closely with the developers who worked on the project, productionising the resulting AI capabilities. The KTP team ensured that the above goals were met by splitting the project into two distinct phases:

Phase One

Rimilia’s clients were separated into different debtor businesses, and their invoices were used to ‘train’ debtor specific computer regression modelling software which allowed the program to predict when future invoice payments would be made. The modelling software accurately predicted (+/-3days) up to 70% of Rimilia’s clients’ customer-base (figures vary between customers). The methodology was rolled-out across multiple Rimilia clients and identified that up to 70% of their customers could be classified as having predictable invoice behaviour. Further supplementary work, at this phase included identifying debtors’ behavioural factors which could impact debtors’ predictability and providing analytical reports on clients’ customer-base performance.

Phase Two

This phase focused on creating a ‘pattern recognition model’, which modelled the weekly payment patterns of the debtors. Note that in phase one the team were trying to predict when invoices would be paid, this phase aimed to predict the total amount each debtor will pay - a debtor may have tens of thousands of invoices for varying amounts. The solution was able to identify the proportion of debtors (figures vary between customers) payment patterns as predictable. The debtors with fully predictable payment patterns and the predictable debtors in terms of invoice payments dates overlap partially.

Ali Yazdi, lead data scientist, (the KTP Associate) said:

“The Knowledge Transfer Partnerships are a great opportunity of academics to get a reality check and re-focus on issues directly of interest to a sector from industry.” In this partnership, we brought together, the latest AI and Machine Learning expertise of university academics and researchers with the real time business expertise of the company to identify and then solve or shed lights on the problems that really matters to the industry. This is what I call Innovation: finding Smarter ways to do important jobs. I am thrilled to see this collaboration has been effective and fruitful.”

The Results:

Increased Brand Exposure

The research completed during this project uncovered how AI could be used to track debtor behaviour accurately. This groundbreaking development was shared at a number of national & international conferences, seminars, webinars, and during both prospect and customers visits. Doing so, increased their exposure and positioned Rimilia as the thought-leaders within their field.

Increased ‘Alloc8 Cash’ Product Sales

Rimilia has used both the product of the KTP (the new forecast engine and capabilities) as well as the insight and knowledge gained from the research, to enhance its products further and drive additional sales. Sales attributed to the KTP project is expected to reach over £1.8m in the next two years.

Upskilled the Rimilia Team

Furthermore, the KTP has addressed the need to increase and embed the Artificial Intelligence (AI), Machine learning (ML) and Data Science (DS) understanding across the company. Over 50 members of the Rimilia team were upskilled to understand AI and ML, and how it helps Rimilia’s solutions with AI being at the centre of their innovative culture.

Supported the introduction of new AI-based solutions

The research, intelligence and analytics gained from the KTP (e.g. around predictability, behavioural fluctuations, debtors’ clustering/segmentation) substantially supported the design and thought development and creation of Rimilia’s new solution called Insight and generation of Financial Relationship Management, a category of solutions first introduced by Rimilia in November 2019.