How Structured Data Enables AI in Finance
This is an excerpt from a presentation called “The CFO’s Guide to AI Strategy,” delivered live at the Intacct Advantage conference on October 19, 2017. Scroll down to see the full slide deck. Click here to download the full white paper.
Like many finance professionals, we at Abacus are skeptical of artificial intelligence.
Don’t get me wrong. Working on machine learning is an important part of what we do. But even more important is how we structure data and make it usable.
We approach AI as the solution with caution. No matter what problem we’re trying to solve, we always try to break it down to a manageable size and lead with the simplest solution.
In terms of AI, that philosophy has led us to do two things: to enable automation by making granular expense data accessible, and to apply that automation in targeted parts of the expense workflow, speeding up the entire process. The value derived from our approach to AI is not to replace teams, but to make them as efficient as possible.
Giving customers access to their expense data has always been a guiding principle for our product. When we started Abacus in 2013, we could see that batched, end-of-month expense reports used by many organizations were hiding expense data and preventing real analysis.
We figured that by eliminating the need for employees to create expense reports at all, our customers could reorganize the expense data much more effectively. Flipping that workflow on its head, finance teams would be able to review reports that were actually meaningful and employees could submit expenses far more efficiently.
Over time, we were happy to see those assumptions bear out. Freeing up expense data also allowed Abacus to automate other parts of the process. For example: given a set of historical spending behavior and data from a card feed, we could use machine learning to predict, with a high degree of reliability, the correct expense category for a given transaction. It allowed us to present relevant options to employees, at the exact right moment, which both sped up expense submissions and made them more accurate.
Seeing those results convinced us to continue developing a targeted approach to automation. We realized that asking software to take over every aspect of a complex workflow like expense reporting would be a massive, unnecessary undertaking.
Much more impactful was to break down the workflow into manageable segments, automate the parts that made sense, and let people take on the remainder of the tasks. Augmenting human thinking with machine learning sped up the entire expense workflow.
We believe this philosophy of human / machine collaboration applies generally across financial software. Automation doesn’t represent an outsourcing of decision making, but a sophistication of the process. Arming yourself with cutting-edge processes yields efficiency, insights, and visibility that you can’t get otherwise. Best of all, you can start investing in these capabilities today.
Whenever I come across finance professionals who haven’t gotten around to thinking about AI, it’s typically because they still see it as abstract; something far away.
I’m here to tell you that it’s not. Automation used tactically, in the right situations, removes friction and amplifies your team’s ability to get work done more efficiently, more accurately, and more completely. Your employees will be more productive, and better yet, they’ll create real value for the whole business.
Forward-looking finance teams are not only already benefiting from the targeted use of machine learning and data analysis, but they’re increasingly finding it to be a strategic imperative to maintain their functional strategic advantage over competition.
Here’s how to be tactical, and drive ROI, with your investments in artificial intelligence.