Earlier this year, I sat down with R “Ray” Wang, CEO and Founder of Constellation Research, to get a glimpse of what the future holds for the finance back office.
If you’ve ever been in a room with Ray, you know that he is an engaging and energetic guy. A self-described “futurist,” Ray gets extremely excited to share his vision of the future. It’s almost like he’s been to the future and he came back to tell us what it’s like.
Thanks for spending time with us today, Ray. Let's talk about the Roaring Twenties and the self-driving autonomous enterprise. Where do you see automation, artificial intelligence, and machine learning taking corporate finance and the organization in the future?
The number one question we get from every CFO is, "When can I get to a self-driving autonomous enterprise?" They want to get to a point where there are no mistakes being made. They want to get to the point where applications are self-aware. They want to get to a point where they're self-healing, self-correcting.
This autonomous enterprise ties back to the other questions every company's going to be asking for the next decade.
Now, those two in the middle are really important because when we take a machine and we pair it up with a human, what we're really trying to do is understand nuance. It takes time.
It takes time to get to precision decisions. It takes time to trust the decision and when we pair up a human with a machine, and we're trying to improve the speed, right? We're trying to improve the accuracy.
Now, the real question is “If you don't train humans and machines at the same time, will we lose intuition? Will we miss something?”
And those are the most important questions that we're seeing in the autonomous enterprise.
As a human being do I need to be concerned about the machines taking my job? If I show up on Monday morning, will I not have a job in the finance or accounting department because of robotic process automation or machine learning or artificial intelligence?
Great question. People always wonder, "Are jobs going to be gone if we go full autonomous?" The short answer is no. We've got to train the machines, so you’ve still got to know what you're doing. The question is, do you need as many people to train the machines? Maybe not.
But guess what? That means you get to do planning and budgeting, that you get to do strategic things in the finance department. And you don't get blamed for any errors and mistakes that are going to pop up. Because when you're looking at risk mitigation, a lot of the things that happened are human errors.
"I forgot to move a cell," "I screwed up a macro," "Hey, I actually didn't enter that number." Those are the things that should have been automated, right?
We're trying to reduce risk. We're trying to get to a point where compliance is something you don't think about.
You want to also make sure that we have humans around because we want to be able to go out and break the rules. Humans are really good at making the rules and breaking the rules, and that's where the testing is going to happen. That's where the penetration testing, that's where the security becomes a new type of job that pops up when we talk about things like AI, SecDevOps.
In your opinion, is it taking too long to get where we are? Are we moving too fast? Are we right where we're supposed to be at the right time? And what are the factors that caused us to be here right here, right now where we're talking about artificial intelligence and machine learning in the context of accounting?
Here's a great question thinking about what's happening. Are we moving fast enough in this space? Are we moving fast enough in this industry? Does AI play a role here?
The challenge is this, right? We've been talking about AI forever. We've gone through AI winters. We finally got to this point. The real difference now is the fact that we have the cloud. The cloud solves the storage problem. The cloud solves the computing problem, the cloud solves the compression issues that have to do with getting things faster, time-to-market, and scale.
So, once you have that in place, a lot of these algorithms were all built in the '80s and '90s. They just never had a chance to get tested. They've never seen this much data to actually prove what's going to happen. So, this is probably the perfect time where we get to AI and we get to a level of automation.
The second thing that's happening is a lot of these legacy finance systems, I mean, they are 10, 20. Let's say Y2K was what, 20 years ago? These things are still running! People put these in '97, '98. I was an SAP FICO consultant. Think about that—that's my skill set. A lot of the systems we put in are still running!
Now, the question is, "Well, what do I do with this data?" It's stuck in a monolithic ERP. I've got to extract the data. I want to take that data do some analytics against it. I want to make it real-time. I want to connect it to action. And really what I want to get to is a point where I can orchestrate decisions.
Being able to orchestrate those decisions gives me the opportunity to start tying things and starting to automate, right? We're learning about automation, so if I can't get this process automated, then I really can't get to a point where I can apply AI.
Since you're a futurist, Ray, let's talk about the future. Here we are today—20 years past Y2K. Let's jump another 20 years in the future, into 2040. Hopefully, you and I are sitting on a beach drinking piña coladas. What does the finance and accounting function look like in the year 2040?
2040 is very, very different. Finance and accounting are taken to a point where full automation, full compliance is in place. But what we're starting to do is we're putting people in place to think about situational analysis, situational awareness. What's happening? What we're looking for is very, very interesting exceptions and forecasting that's going on.
The data being collected sits in networks of networks. These networks are creating demand signals, supply chain pricing optimization. They're giving you cues in terms of what customer wants are. And the finance is basically the brains of every operation because it's trying to understand where optimization occurs.
So, analytics and predictive analytics, we used to build these charts and these tables. Now that's all done right away. It just gets sucked in. The models are built automatically. The tables are there. People check and test those models, then they deploy it. That's not a one month exercise. That's like a day, right? You may spend a day testing it, modeling it, checking it, and then you go deploy. So, it's much faster, it's much more accurate, and it's much more compliant.
In your opinion, Ray, what types of companies—large, small, private, public, for-profit, non-profit—can benefit from this type of technology in the back office?
Oh, this is a technology for every company. We're not looking at specific industries. But what we're going to see is the highly regulated industries are going to be able to take advantage of this the quickest. Why? Because regulation gives you the guide rails. It gives you the rules upfront and you can build models very quickly there.
But once that's in place, the learnings that are put there will help a small business actually get started. They'll get the same impact that a larger enterprise will have by plugging into a network, by taking the learnings, by building onto the models, by subscribing to the algorithmic libraries that are there. That gives them the same type of advantage.
The only difference between companies that are not on an AI-based type of automated self-driving autonomous solution and the ones that are is that people are going to spend more time on the things that really don't matter, right?
Things that we really worry about that aren't important. So, if you remember Maslow's hierarchy of needs, you start with the ego, you worry about your safety and security on the bottom. Well, the same thing happens in the enterprise. There’s an enterprise version of Maslow’s hierarchy.
The bottom rung is regulatory compliance. Don't get me fired, don't get me sued, don't kill anybody. The operational efficiency is the next rung, which is for every dollar invested, it's going to save you one, two, or three. The next rung is revenue and growth. For every dollar invested you're going to make two, three, or four. And then we have strategic differentiation. And then we have your brand, which is the ego of your company, it's like a person. Hey, who would they be?
Well, the problem is every company optimizes for operational efficiency and regulatory compliance. That's where all the resources go. That's where all the budgets are. And that's how they staff.
Well, that's the problem why companies are failing. You can't grow if everything's set up like this. You want to flip the pyramid. Focus on your brand, your purpose, your mission, go build out the new next business model.
Maybe it's subscriptions, maybe it's something else. And start focusing on revenue and growth. The stuff on the operational efficiency and regulatory compliance—that's got to be automated. That's got to be put into AI. That's got to be software driven. Because if you're not doing that, your best people are working on the lowest areas that you want to be in.
I don't ever want to put a person on that!
Excellent. I love that answer. Hopefully, we're self-actualized in the finance function.
I don't know. We’ll see. (laughs)
What are the career opportunities for those who first embrace artificial intelligence within finance? The innovators: what's this going to allow them to do with their careers?
So early adopters of this technology are going to have a massive advantage. It's really about how you train these systems, how you create new models. You'll be the pioneers. Really thinking about what the future finance looks like, what gets automated. When you think about when humans make decisions, how you become more strategic and how we get to, ultimately, a world of situational awareness.
Now, that's not going to happen right away. It's going to take three years, five years, ten years to get to that point. But the knowledge that you accumulate, being first is going to be something that sets you apart from everyone else. Just like with the beginning of the ERP revolution when we actually went from mainframe to client-server, and then to the internet. That's the same level of shift we're talking about here, but it's 10 times the difference.
You were talking about ERP vendors a moment ago. Why do you think the ERP vendors haven't moved faster on this type of technology?
A lot of people ask about why existing vendors aren't moving as quickly on the technology. It comes to one thing: it's inertia, and it's really the legacy client base. It's very, very hard to get people to move there and the investment that's required to put the R&D around the AI. It takes a lot of effort.
A lot of these companies have spent a long time acquiring companies. They've taken your maintenance dollars and pretty much use that to acquire additional capabilities. And then instead of investing in back into the core product that you paid for, they went and bought something else and then asks you to pay for it. That's the biggest challenge why legacy or big companies can't innovate as quickly.
So when companies are thinking about implementing machine learning, artificial intelligence, natural language processing in the back office, what are the functions that are the easiest places to start, the lowest hanging fruit?
Let's take a thing like invoice matching in accounts receivable. People are doing that manually. Like, "Oh, here's the number, here's the check. Does that make sense?"
That's crazy, right? It's two-way matching, three-way matching. That should happen automatically. We should be able to take OCR computer vision, put that into place, I'll assign some logic so that that's something you never have to do again so you don't miss a payment.
But once we add AI, we get something very interesting. I can figure out, "Hey, that was a 30-day term versus a 60-day term."
And that creates a very, very different dynamic.
On a 30-day term, "Well, hey, maybe if I pay earlier, I lose out. But on a 60-day term, maybe I pay earlier, I might actually have a discount for paying the vendor earlier."
Or maybe I'm just going to pay late because my cash reserves are in a different situation and I'm reporting at the end of the quarter, right? Gives you a lot more choices. It becomes a lot more intelligent. In aggregate, what that does, it gives us the ability to not spend time on the mundane and be more strategic.
You just alluded to cash flow. How do you see artificial intelligence, machine learning, sophisticated datasets, and complex algorithms improving the ability of companies to forecast and manage cash in the future?
When we get to self-driving, self-automated apps, what we're going to be able to do is manage cash. We can go from weekly close to daily close to hourly close. It might even be instant close. "Hey, what's the cash balance?" "Oh well, this is what it is at this moment in time at this second." Right?
It's almost like quantum. Everything changes, the answer is different every hour, every second. Well, it's going to be like that. Now, are you going to really do that? No, you'll go insane. But the fact that you can do that is interesting. You would know exactly how much cash is on hand. You know exactly how much cash you have to deploy in the next 30 days. It completely changes your treasury function.
Awesome. Any concluding thoughts, Ray?
The next decade is going to be something very different, unlike anything we've seen before. We've had transactional systems for so long. Companies that have been in the ERP space have been focused on managing a transaction, working on a business process. Well, that's not as important.
The future is really about getting to precision decisions. And precision decisions happen when you actually automate your transactions, you deploy artificial intelligence, and you can create an autonomous enterprise. That's where we're headed in the next 10 years.
I can’t wait. Thanks so much for sitting down with us today, Ray. It was enlightening.
Always a pleasure. Thanks for having me.