©Jolygon/iStock/Getty Images Plus
FEI Daily spoke with Prophix president Alok Ajmera about the risks of failing to adopt artificial technology (AI) and the problems the technology can (and cannot) solve.
FEI Daily: How will AI and machine learning change the office of finance in 2020?
Alok Ajmera: Let me take a step back. Often, when I'm talking to CFOs, the image that people have is this sentient being. It's the Jarvis of Iron Man, right? And that is a longer-term aspiration. I think the thing that needs to be said is AI is an umbrella term, encompassing a whole bunch of different types of technologies and you can categorize them into two buckets.
Up until now, the way most people, especially in the corporate finance group, are interacting with technology, you're kind of hunched over a computer, plugging away with your mouse and the keyboard and staring up at a screen and then all of a sudden you're getting neck pain and carpal tunnel because it's not a very natural way to work. So, the first bucket of AI is all about the humanization of technology.
The other bucket is actually the complete opposite. It's the making it more machine-like. To doing things or solving problems in a way that is not natural for humans, but more natural for a machine. What's really exciting about machine learning today is that there are now specific problems that can be solved in the office of finance using machine learning. The most natural one is around anomaly detection. If you think about what accountants are doing on a month to month basis or quarter to quarter, when you're closing books, et cetera, it's really about looking at large sets of data and applying business logic to look for errors, mistakes, fraudulent transactions, et cetera. And that's a use case that's really well-suited for machine learning. Learning patterns of your data, applying business logic, and then catching exceptions. That's a real problem that people can solve with machine learning in 2020.
FEI Daily: What does an organization risk if they don’t harness AI this upcoming year?
Ajmera: There is a massive wave of digitization that's occurring and it's already occurred across big chunks of the organization and the enterprise. It's just now catching up with the office of finance. Oftentimes we'll speak to financial executives and they're really excited about AI and machine learning, but at the same time, a lot of their processes are incredibly manual. They're not well-structured, they're not using technology today.
I think the risks go beyond just AI. Digitization is coming. Making better use of the data that your organization has to make better decisions is coming, it's happening and people are doing it. And if you're not going to follow suit, you're going to be left behind.
If you approach AI or technology as a means to just streamline the work that you're doing, I think you're going to miss the mark. If you realize that you can use these types of technologies, AI, et cetera, to harness the data that you have, to make better decisions, to be more forward-looking from a business perspective, to be more proactive in the business, you're going to be driving a competitive advantage. And if you're not going to take advantage of that, then your competitors will, and that's the bigger risk, right?
FEI Daily: What are some initial steps finance teams can take so they don’t fall behind?
Ajmera: It depends to a certain degree where they are in terms of their current digitization. If they're further behind and they have a lot of manual processes, a lot of one off Excel spreadsheets floating around, the first thing they need to do is start creating some structure and process and infrastructure in place to actually automate some of the mundane things that they're doing so that they have time and energy to reinvest in more forward-looking technologies.
If they are already down that path and they've found ways to streamline and automate with technology, a couple of things that might be useful for these companies: start looking at the skillset that the team has. Take machine learning as an example. This is not the same skillset as hiring an F&A person. You might require totally different skills, totally different types of people. You might be looking at a data scientists. And, so, look at your teams and be a little more critical of the skills that you have or do not have. You might already have analysts that are incredibly quantitative, and supporting them on some training and learning could actually help augment that skill. There's really no point going down this path without assembling the right people, right skills, which will change dramatically.
On the other end, there's technologies and technology partners that they should already be talking to. Not ‘I need to go out and hire an engineering team, look at some raw AI platforms and start building something.’ It’s more about looking for the right business partners that are using these types of technologies.
The last thing I would say, is that it's really easy to get swept up in the buzz and excitement of AI. A lot of companies are spending energy, creating strategies and doing stuff, but not actually getting any value at the end. I think the key is look for real problems that can be felt and if that problem does not exist in your organization or solving it is not going to add any value, then move onto something else. Don't just do this for the sake of doing it.
A good example, we have a customer who was really excited about machine learning and they wanted to partner with us around using the data that they have and creating machine learning algorithms to help forecast in a more accurate way. We said, ‘yeah, we can definitely help you with this.’ We started working with them only to realize that the data that they had was a disaster. They didn't have the depth and the structure of the data that they would be required to create meaningful forecasting algorithms. And, so, we had to stop and we're like, ‘I know it's exciting and I want to sell you all this cool machine learning technology. However, before we even get there, you need to spend some time putting some disciplines together around the data that you have, especially on the machine learning side.’
It's all about quantity, quality, depth, and breadth of data. And a lot of organizations struggle with that.