QAD Explore, Prophix, data, machine learning

Our second featured Explore article is prepared by our friends at Prophix. Prophix develops innovative Corporate Performance Management (CPM) software that automates critical financial processes such as budgeting, planning, consolidation and reporting – improving a company’s profitability and minimizing their risk.

Machine learning has seemingly unlimited potential to make your Finance department more efficient and effective. But it can be daunting to consider making the changes required to take full advantage of the technology.

Most companies nowadays are taking advantage of technology solutions, and the financial market is not exempt from that. For example, those wanting to start a brokerage will use resources online to find Forex leads, as well as discover other essential information from the Web and different types of software. Technology has changed these markets for the better.

“The prospect of artificial intelligence is going to be very intimidating on multiple levels,” Jack McCullough, President of CFO Leadership Council, told us in a recent interview. “While the promise of AI is amazing, we are still in its infancy to a degree…[but] embrace it as a positive! It’s going to improve your company. It’s going to fundamentally improve the very quality of life for people all over the world. And that’s a fact.”

The question on most Finance leaders’ minds is: Where do I start?

The good news is that applications of machine learning in Finance isn’t all or nothing. The key is to start with a project that is both low-risk and high-impact. This allows you to secure an easy win and build from there.

With that said, here is a simple roadmap for adding machine learning into your finance processes, along with current applications.

Step 1: Start with Automation

Machines love routine tasks. People? Not so much. Finance automation has the dual benefit of being the easiest machine learning for Finance groups to implement, and the one with the most tangible results.

The first step is identifying a project that is critical, but also time-consuming, repetitive and data-dependent. Of course, for the automation to yield excellent results, the input data and data pipeline need to be in good shape too. So, make sure that data is sanitized and well-structured, and that relevant parties are trained on data management and hygiene before initiating any project.

The following functions are prime targets for machine learning. By automating these common tasks through a single platform, organizations should experience a solid win for machine learning.

Procure-to-pay (P2P): An automated P2P process can provide Finance leaders improved high-level visibility on organizational spend as well as the day-to-day minutiae of invoice and PO status, arming them with facts they need for timely accruals and query resolutions. Typical steps in an automated P2P process include requisition, invoice capture, invoice matching, invoice template approval and ERP integration.

Order-to-cash: In the traditional order-to-cash process, different business functions use their own systems and data, resulting in inefficient processes and imperfect data. By automating the order-to-cash process, Finance groups can expect increased awareness of risk ratings, quicker turn on financial documents and more accurate invoicing – all of which improve cash flow and efficiency. In fact, data from the IBM Institute for Business Value suggests that improving the order-to-cash practices can lead to an 83% improvement in performance.

Record-to-report (R2R): Automating the R2R process can deliver a faster financial close, improve business compliance, help ensure the integrity of financial reporting, and provide continuous monitoring of KPIs and flash reporting. Milestones in an integrated R2R process include: assimilation of data from subledger entries, integration of data into the general ledger, aggregation of the data and automated reporting.

Step 2: Ramp up with Augmentation

After offloading routine tasks to automation, the question becomes: How can machine learning help Finance organizations make better decisions? That’s the basis of augmented intelligence, and it’s already playing a key role in many Finance organizations through:

Fraud Reduction and Security: Finance has long depended on the processing power of computers to identify anomalous behavior. The difference is that while previous systems were the product of a complex and robust set of rules, newer systems actively learn and adapt based on perceived security risk. As a result, possible fraud and security issues are flagged sooner. John Colthart, VP of Growth at MindBridge AI, explains this relationship further:

“AI catches the errors and the anomalies, the potential for issues that an auditor, an accountant or financial professional would then investigate. When they do that investigation with the right information from the AI, they’re able to be more specific in their questioning, and they’re going to be able to find intent. The minute they find intent, they’re going to be able to claim that as something fraudulent.”

Data Management: Augmented AI helps overcome challenges with internal data management, bringing together disparate data and highlighting insights to shape business decisions. Another application for augmented AI is sorting through hundreds of thousands of emails or form submissions, determining priority communications and the sender’s intent.

Customer Service: Using augmented AI can help Finance organizations remain customer-focused as consumer expectations for service and responsiveness increase. For customer-facing Finance groups, chatbots and conversational interfaces are seen as having huge potential. Some organizations are also utilizing augmented robo-advisors who can provide detailed answers about savings and loans.

Step 3: Bring in AI for Analytics & Prediction

The next level of machine learning is leveraging computer processing power to analyze data, quickly make assumptions, perform a scenario analysis and predict outcomes. AI systems can review up to 800 million pages of text per second and even ingest new regulations as they are created, ensuring that their assumptions are always spot-on.

Current applications of data-driven decision-making include hedge fund management and algorithmic trading. The effects of machine learning taking an advisory role can be seen in the transformation of the underwriting process. However, much of the potential for AI-driven predictive analysis is untapped and Finance Leaders can expect other developments in the future.

Getting Started with Machine Learning

Machine learning has the power to transform the business of Finance by automating routine tasks, augmenting human decision-making, and accurately predicting outcomes. But, tapping into the power of machine learning is an ongoing process. The important thing is to start.


  1. That’s a great article, thank you for this valuable information. I think Machine Learning technology is perfect for the finance industry. Automation is probably my favorite way to use ML. However, it could be highly effective in Analytics and Prediction too. Speaking of Augmented Intelligence, I think this implementation still have a room to grow, but current use cases are still impressive. Every business in Finance, or any other industry should be aware to Machine Learning and plan to implement it to the processes!

  2. Thanks for the article! There’s been a variety of machine learning-based methods proposed, both supervised and unsupervised, to tackle the issue of fraud detection. The supervised approaches rely on explicit transaction labels i.e. machines need to be shown, repeatedly, what genuine transactions look like during training to be able to distinguish the fraudulent ones later.