theCUBE is the world’s leading live interview show covering enterprise tech, innovation and the people who imagine, create and implement the technologies that are changing our world. This year, they joined IFS in Boston for IFS World Conference 2019. Below is the transcription of an interview with Director of IFS Labs, Bas de Vos, and IFS Vice President of AI and RPA, Bob De Caux. Enjoy!
theCUBE: We’re back in Boston, Massachusetts, IFS World, day one. You’re watching The Cube, Dave Vellante with Paul Gillin. Bas de Vos is here, he’s the director of IFS Labs, and Bob de Caux, who’s the vice-president of AI and RPA at IFS. Gents, welcome, good to see you again.
BdV: Good morning.
theCUBE: Bas, you were on last year, talking about innovation, IFS Labs. First of all, tell us about IFS Labs and what you’ve been up to in the last twelve months.
BdV: Well, IFS Labs functions as the new technology incubator for IFS right, so we’re continuously looking at opportunities to bring innovation into product and help our customers take advantage of all the new things out there to yeah, to create better businesses. And one of the things I talked about last year is how we want to be close to our customers, and I think that’s what we have been doing over the past year, really being close to our customers.
theCUBE: So, Bob, you got the cool title, (Bob laughs) AI, RPA, all the hot cool topics.
BDC: Yeah, all the good names.
theCUBE: So, help us understand what role you guys play as IFS as a software developer. Are you building AI? Are you building RPA? Are you integrating it? Yes, yes? Paint a picture for us.
BDC: Yeah, I mean, our value to our customers comes from wrapping up the technology, the AI, the RPA, the IoT, into the product in a way that’s going to help their business. So it’s going to be easy to use, they’re not going to need to get a technical specialist to take advantage of it. It’s going to be embedded in the product in a way they can take advantage of very easily. That’s the key for us as a software developer. We don’t want to offer them a platform that they can just go and do their own thing, we want to sort of control it, make it easier for them.
theCUBE: So I presume it’s not a coincidence that you guys are on together. So this stuff starts in the labs, and then your job is to commercialize it, right? So, so take machine intelligence, for example, I mean it can be so many things to so many different people. Take us back to sort of, you know, the starting point, you know, within reason, of your work on machine intelligence, what you were thinking at the time, maybe some of the experiments that you did, and how it ends up in a product.
BdV: Well, very good question, right. So I think we start at, well, first of all, I think IFS has been using machine learning at various points in our products for many many years. So, for example, in our dynamic scheduling engine, we have been using neural networks to optimize fuel service scheduling for quite so many years. But I think if we go back like two years, what we saw is that there’s a real potential in our products that, if you would take machine learning algorithms inside of the product, to actually help automate certain decisions in there, that it could potentially help our business quite a bit. And the role of IFS Labs back in the day is that we just started experimenting, right, so we went out to different customers, we started engaging with them to see, okay, what kind of data do we have, what kind of use-cases are there? And basically based on that we sort of developed a vision around AI, and that vision, back in the day, was based on three important aspects: human-machine interaction, optimization, and automation. And that kind of really landed well with our customer use case. We talked quite a bit about that at the previous World Conference. So, at that point, we basically decided okay, you know what, we need to make serious work of this. Experimenting is good, but at a certain point, you have to conclude that the experiments are successful, which we did, and at that point, we decided to look at okay, how can we make this into a product? And how that normally goes is that we start engaging with them more intensively and starting to hand over. In this case, we decided it was also a good moment to bring somebody on board that actually has even more experience and knowledge in AI than what we already had as IFS Labs, but that could basically take over the baton and say okay, now I am going to run with it and actually start commercializing and productizing that. Still in collaboration with IFS Labs, but yeah, taking that next step in the road and then Bob came onboard.
theCUBE: Christian Pedersen made the point during the keynote this morning that you have to avoid the appeal of technology for technology’s sake. You have to have, it has to start with the business-use case. You’re both very technical, very deep into the technology, how do you keep disciplined to avoid letting the technology lead your activities?
BdV: Well, may I Bob?
BDC: Yeah, absolutely.
BdV: Yeah, so I think good examples will be seen at this World Conference as well. It is staying close to the customer, and accepting and realizing that there is no, there is no use in just creating technology for the sake of technology, as you say yourself. So what we did here, for example, is that we showcase collaboration for checks with customers. So, for example, we showcase one with Cheer Pack, which is manufacturing spouted pouches, down here in Massachusetts actually, and they wanted to invest in robotics together with us, so what we basically did is actually went into their factory, literally on the factory floor, and start innovating there. So, instead of just thinking about how do robotics and IFS applications, or one of our other products, work together? We said, let’s experiment on the shop floor of a customer, instead of inside of the ivory tower, as sometimes our competitors do it. Does that answer your question?
theCUBE: Yeah I think it does.
BDC: I can pick up a little.
theCUBE: Yeah I’d love to see some other examples too.
theCUBE: Well, so I think the really important thing, and again, Christian touched on it this morning, it’s not the individual technologies themselves, it’s how they work together. We see a lot of the underlying technologies becoming more commoditized, that’s not where companies are really starting to differentiate. Algorithms, after a while, become algorithms. There’s a good way of doing things. They might evolve slightly over time, but effectively you can open source a lot of these things, you can take advantage. The value comes from that next layer up, how you tape those technologies together, how you can create end-to-end processes. So if we take something like predictive maintenance, we would have an asset, we would have sensors on that asset, that would be providing realtime data to an IoT system. We can combine that with historical maintenance data stored within a classic ERP system. We can pull that together, use machine learning on it to make a prediction for when that machine is going to break down, and based on that prediction we can raise a work order. And if we do that over enough assets, we can then optimize our technicians so, instead of having to wait for it to break down, we can know in advance, we can plan for people to be in the right place. It’s that end-to-end process that’s where the value is. We have to bring that together in a way that we can offer it to our customers.
theCUBE: There’s certainly, you know, a lot of talk in the press about machines replacing humans. Machines have always replaced humans, but for the first time in history, it’s with cognitive functions now, so people get freaked out a little bit about that. I’m hearing a theme of augmentation, you know, at this event, but I wonder if you could share your thoughts with regard to things like AI automation, robotic process automation. How are customers, you know, adopting them? Is there, sort of, concern upfront? I mean, we’ve talked to a number of our PA customers that, you know, initially maybe are hesitant, but then say, wow, I’m automating all those tasks that I hate. And then you sort of lean in. But at the same time, you know, it’s clear that this could have an effect on people’s jobs and lives. What are your thoughts?
BDC: Sure, do you want to kick off on that?
BdV: Yeah, well no, if you can.
BDC: Yeah, absolutely, that’s fine. So I think, in terms of the automation of the low-level tasks, as you say, that can free up people to focus on high-value activities, something like RPA, those bots, they can work 24/7, they can do it error-free, it’s often doing work that people don’t enjoy anyway. So that tends to actually raise morale, raise productivity, and allow you to do tasks faster. The augmentation I think is where it gets very interesting because you need to, you often don’t want to automate all your decisions. You want people to have the final say, but you want to provide them more information, better, more pertinent, ways of making that decision. And so it’s very important, if you can do that, that you’ve got to build trust with them. If you’re going to give them an AI decision that’s just out of a black box, and just say there’s a 70% chance of this happening, you know, what I’ve found in my career, is that they don’t tend to believe that, or they start questioning it, and that’s where you have difficulties. So this is where explainable AI comes in. However, being able to state clearly why that prediction’s being made, what are the key drivers going into it, or if that’s not possible, at least giving them the confidence to see well, you’re not sure about this prediction, you can play around with it, you can see that I’m right, but I’m going to make you more comfortable and then hopefully you’re going to understand and sort of move with it. And then it starts sort of finding it’s way more naturally into the workplace. So that’s, I think, the key to building up a successful organization.
theCUBE: So is that essentially what it is? It’s sort of giving a human the parameters, the probabilities, and saying, okay, now you can make the call as to whether or not you want to place that bet, or make a different decision, or hold off and get more data. Is that right?
BDC: Yeah, I think a lot of it is about setting, you know, the thresholds and the parameters within which you want to operate. Often if a model is very confident, either, you know, a yes or a no, you’ll probably be quite happy to let it automate and take that through. It’s the borderline decision where it gets interesting. You probably still want someone to look over it, but you want them to do it consistently, you want them to do it using all the information to hand. And so that’s what you would do, you’d present it to them.
BdV: And to add to that, I think we also should not forget, is that a lot of our customers, a lot of companies, are actually struggling finding quality staff right. I mean, aging of the workforce, right, we’re all retiring eventually, right? So aging of the workforce is a potential issue. Finding lack of quality staff. So if I go back to the Cheer Pack example I was just talking about, and some of the benefits they get out of that robotics project, is, of course, they’re saving money, right, they’re saving about 1.5 million dollars a year on money on that project, but their most important benefit for them is actually the fact that they have been able to move the people from the work floor doing that into higher-skilled positions, effectively countering their labor shortage. They were limited in their operations by the fact they had too few quality staff. And by putting the robots in, they were able to reposition those people. And that’s, for them, the most important benefit. So I think there’s always a little bit of a balance, but I also think we eventually need robots, we need automation to also keep up with the work that needs to be done.
theCUBE: Maybe you can speak to, Bob you can speak to software robots, when people think of robots they tend to think of machines, but in fact software robots are where the real growth is, right now, the greatest growth is right now. How pervasive will software robots be in the workplace, do you think, in three to five years?
BDC: I think the software robots, as they are now, within the RPA space, they fulfill a sort of part of the overall automation picture, but they’re never going to be the whole thing. I see them very much as bringing different systems together, moving data between systems, allowing them to interact more effectively. But within systems themselves, you know, the bots can only really scratch the surface. They’re interacting with software in the same way a human would, on the whole by clicking buttons, going through it, etc. Beneath the surface, you know, for example within the IFS product, we have got data understanding how people interact with our products, we can use machine learning on that data, to learn, to make recommendations, to do things that a software bot wouldn’t be able to see. So I think as a combination, you know, the software bots are kind of on the outside looking in, but they’re very good at bringing things together, and then inside you’ve got that sort of deeper automation to, you know, take real advantage of the individual pieces of software.
theCUBE: This may be a little out there, but you guys are deep into the next generation. I want to talk right now about quantum and how we could see workable quantum computers within the next two to three years. What do you think the outlook is there? How is that going to shake things up?
BdV: Yeah, it’s, let me answer this. We’re actually having an active project in IFS Labs currently, looking at quantum computing, right. There’s a lot of promise in it, there’s also a lot of unfulfilled promise in there, right. But if you look at the potential, I think where it really starts playing into benefits is if, the larger the optimization problems, the larger the algorithms are that we have to run, the more benefit it actually starts bringing us. So, if you ask me for an outlook, I say there is potential, definitely, especially in optimization problems, right, but I also think that the realistic outlook is quite far out. Yes, we’re all experimenting in it, and I think it’s our responsibility as IFS, as IFS Labs to have a look and what it could potentially mean for applications as we have as IFS, but my personal opinion is the outlook is, yeah, at least five to ten years out, right.
theCUBE: What comes first, quantum computing or fully autonomous driverless vehicles?
BdV: Oh, that’s a tricky question.
BDC: I mean I would say, in terms of the practical commercial application, it’s going to be the latter. I mean they’re much closer
theCUBE: Wow, okay.
BDC: to pulling that off.
theCUBE: So that’s quite a ways off.
BDC: I think so.
theCUBE: A question back on RPA, what are you guys exactly doing on RPA? Are you developing your own robotic process automation software, are you integrating? Doing both?
BDC: So, within the product we, you know, if we think of RPA as this means of interacting with the graphical user interface in the way that a human would, within the product we’re thinking more in terms of automating processes using the machine learning, as I mentioned, to learn from experience, etc., in a way that would take advantage of things like our open API’s, that were discussed on main stage today. RPA is very much our way of interacting with other systems, so allowing other systems to interact with IFS, allowing us to send messages out. So we need to make it as easy as possible for those bots to call us, you know, that can be by making our screens nice and accessible and easy to use. But I think the way that RPA is going, a lot of the major vendors are becoming orchestrators really, they’re creating these studios where you can drag and drop different components in to do OCR, provide cognitive services and, you know, elements that you could drag and drop in would be to say take data from a file and load it into IFS and put it in a purchase order. And you could just drag that in. And then it doesn’t really matter how it connects to IFS, it can do that via the API, and I think it probably will. So, it’s creating the ability to talk to IFS that’s the most important thing for us.
theCUBE: So you’re making your products RPA ready, friendly.
BDC: Exactly.
theCUBE: It sounds like you’re using it for your own purposes, but you’re not an RPA vendor per se, you’re not saying, Okay, here’s how you do an automation.nYou’re going to integrate that with other RPA leadership products.
BdV: I think we would really take more of a partner approach there, right, so if a customer, I mean there’s different ways of integrating systems together. RPA is a good one there, there’s other ways as well, right. That if a customer actually wants to integrate their systems together using RPA, very good choice. We make sure that our products are as ready as much for that as possible. Of course, we will look at the partner ecosystem to make sure that we have sufficient and the right partners in there that the customer has a choice in what we recommend. But basically, we say we want to be agnostic to what kind of RPA vendor sits in there.
theCUBE: Notwithstanding, there’s obviously a lot of geopolitical stuff going on with tariffs and the like, so notwithstanding that, do you feel as though things like automation, RPA, AI, will swing the pendulum back to on-shore manufacturing, whether it’s Europe or U.S., or is the cost still so dramatically advantageous to, you know, manufacture in China? Will that pendulum swing, in your opinion, as a result of automation?
BdV: Good question. I’m not sure it will completely swing, but it will definitely be influenced, right. One of the examples I’ve seen in the RPA spaces right, where a company before would actually have an outsourcing project, in India, where people would just type over the purchase orders right? Now an RPA bot scans it in, so they don’t need the Indian offshore anymore. But it’s always a balance between, you know, what’s the benefit, what’s the cost of developing technology and it’s almost like a macro economical sort of discussion. One of the discussions I had with my colleagues in Sri Lanka and maybe a completely off-topic example, we were talking about car wash, right? So us, in the Western world, we have a car wash where you drive your car through, right? They don’t have them in Sri Lanka, and all the car washes are by hand. But the difference is, because labor is cheaper there, that it’s actually cheaper to have people washing your car while, with us in the U.S., for example, that’s more expensive than actually having a machine doing it, right? So it is a macro economical sort of question. That’s quite interesting to see how that develops over the next couple of years.
theCUBE: All right, yes, well thanks very much for coming on The Cube. Great discussion, really appreciate it.
BdV: Thank you very much.
BDC: Thanks.
theCUBE: All right, you’re welcome. All right, keep it right there everybody, Dave Vellante, Paul Gillin. We’ll be back, IFS World, from Boston, you’re watching The Cube.
Do you have questions or comments?
We’d love to hear them so please leave us a message below.
