Technology

How Machine Learning Can Help You Reconcile Faster

Dec 03, 2020 | By Carlos Avila

work smarter not harder accounting workflow automation

The terms artificial intelligence and machine learning get thrown around a lot these days. In fact, searches for these terms have skyrocketed over the last decade, going from relatively flat to peak popularity in 2020, according to Google Trends.

As a consumer, it is all too easy to get caught in the barrage of marketing and sales pitches that promise to solve all of your problems with the help of AI and ML. And who can blame you? AI and ML are extremely powerful areas in their own right and show nearly unbounded technological potential. Unfortunately, many of these solutions often fall short of their promises and offer little real help to the end users they were built for.

I, for one, am happy to see more honest and intelligent people coming forward to help consumers make better-informed decisions in this regard and offer a trusting hand when navigating this often-unfamiliar landscape. If there is any one key point I hope you take away from this article, it would be this: 

Machine learning is not the goal. Helping you do a great job is the goal.

FloQast employs machine learning to help accountants do a great job by reducing tedious tasks that can be safely automated. In order to do that, we’ve spent the last few years separating the hype from reality and want to arm you, as the consumer, with the right questions to ask when considering solutions.

Defining AI and ML

In order to understand how an AI/ML solution can best help you, we first need to define it so that you can ask the right questions when shopping.

Artificial Intelligence

Artificial intelligence can be loosely defined as something that mimics intelligent behavior. That’s it — something that mimics intelligent behavior. In most cases, this is code, an algorithm, defined rules, or some other assortment of techniques.

If this definition seems oddly broad and somewhat unhelpful, that’s because it kind of is. It hasn’t been too long since TechCrunch compared the use of the term “AI-powered” in software solutions to the use of the term “all-natural” in food products.

Now, don’t get me wrong, using artificial intelligence to automate your most tedious tasks is definitely valuable to you as a user, but the real difference-maker comes down to how you actualize your AI and this is where machine learning comes in.

Machine Learning

Machine learning is a subset of AI that empowers a computer with the ability to learn. That same computer can now take what it’s learned and apply it to solving a new problem that it has never seen before. 

An empowered machine will create the code, algorithm, or rules necessary to achieve AI. One way to think of this is that a machine is creating the thing that helps us, as opposed to, say, a developer writing code or an accountant creating formulaic rules.

How a computer learns is somewhat more complicated than this article will go into, but a visualization that usually helps is to think about the process of learning to walk. When learning to walk, you want to minimize the cost of falling down, but it takes a lot of practice. As you become better at walking, you may encounter new challenges like different terrains or obstacles. Learning how to walk allows you to take on new challenges you may not have tried before, like running, jumping, and so on.

The Bottom Line: Artificial Intelligence vs. Machine Learning

While artificial intelligence can be your coffee maker that starts making coffee at 8 a.m. because of the schedule you set, machine learning can learn to start making coffee later on the weekends based on your behavior of sleeping in.

Why FloQast ML Works for Accountants

FloQast uses AI and ML to help automate tedious tasks for accountants so that they can focus on bigger value projects. We do this by focusing on the following key points:

  1. Help solve the harder problems
  2. Quickly actualize ROI
  3. Do less work, add more value

Let’s quickly dive a bit deeper into each of these to see why they are a challenge and how FloQast is leading the way in solving them.

We Want to Help Accountants Solve Their Hardest Problems

Matching, on the surface, seems like an easy problem to solve. If the data being entered is consistent and items can be found with a vLookup, then perhaps it is easy to match and you are all done. 

In reality, the problem is usually much more complicated. There are thousands of transactions, varying text fields and abbreviations, offset dates from different systems, and inexact amounts. You can have many transactions that match many other transactions. To make matters more complicated, you can have multiple valid matches or introduce minor mistakes as you progressively match through the month.

Not only does AutoRec handle the most simplistic cases of one-to-one, but our engine is able to make sophisticated matches that had previously been extremely difficult to do, and would be unthinkable by hand.

We Want Accountants to Quickly See ROI

The primary challenge with traditional machine learning is that it usually requires lots of data and time. The problem: Time is something most teams don’t have much of. It makes little fiscal sense to buy a solution today and hope that it might pay off a few years later. Accountants need to see payback quickly. This is something that FloQast already prides itself on

AutoRec offers high-quality, out-of-the-box results. What that means is that from day one, you will see value. Over time, that value will build on itself as AutoRec learns more about the nuances of your data. 

For years now our clients have leveraged our ML-based matching engine, AutoRec. We have been iterating on our solution for a few years, and in that time, we’ve learned a great deal about how to best approach a client's transaction data.

We Want Accountants to Do Less Work, While Adding More Value

Swapping out tedious work for other types of tedious work is not a good tradeoff. When encountering a problem, such as matching transactions, accountants often resort to complex rule systems, detailed configurations, and, when those don’t work, brute force. The problem with these approaches is that they are brittle, place an unnecessary administrative burden on the overall process, and require you to allocate a good chunk of your mental capacity to maintain all the little details month over month, which can lead to costly mistakes.

AutoRec is a no rules, no config, and easy-to-setup solution. This means you can leave the tedious work to FloQast and count on it to find the best way to work with your data. There is no overhead in maintaining this process and we will be able to keep up even as your company changes.

To the Future

Artificial Intelligence and Machine Learning will fundamentally change the way accountants approach their work. The good news is that automation isn’t here to take away jobs. AI and ML will instead help accountants do their best work while further expanding what is possible in the role. We hope the above helps you better understand what you should be looking for in an ML solution. Please feel free to reach out to a FloQast rep to learn more about what FloQast and Machine Learning can do for you.

Carlos Avila
Carlos Avila is an Engineering Manager at FloQast focused on developing top talent and a world class AutoRec product. Previously a Technical Leader on FQ Close, Flux, and Analyze, Carlos considers himself a recovering full-stack-aholic and craftsman of all things product engineering.

Check out research, videos, case studies, and more!

Learn more about working at FloQast!