HRx Radio – Executive Conversations: On Friday mornings, John Sumser interviews key executives from around the industry. The conversation covers what makes the executive tick and what makes their company great.

HRx Radio – Executive Conversations

Guest: Mike Gioja SVP of IT and Product Development, Paychex
Episode: 349
Air Date: December 20, 2019

Transcript

Important: Our transcripts at HRExaminer are AI-powered (and fairly accurate) but there are still instances where the robots get confused and make errors. Please expect some inaccuracies as you read through the text of this conversation. Thank you for your understanding.

Full Transcript with timecode

John Sumser 0:13
Good morning and welcome to HR Examiner’s Executive conversations. I’m your host, John Sumser and today we’re going to be talking with Mike Gioja, who is the Paychex Senior Vice President of IT and product development. Now, you wouldn’t necessarily be sitting in your car listening to this going, Oh, Paychex, those leaders in artificial intelligence and HR tech, but I think what you’ll see where we talked to Mike is that the company has a deep and interesting story about how they’re approaching new technology. Good morning, Mike, how are you?

Mike Gioja 0:47
Good morning. How are you?

John Sumser 0:48
I’m great. I’m great. I’m looking forward to this conversation. So take a minute and introduce yourself and be sure to talk a little bit about your time in the industry. You’ve been at this a while and so please take a moment and introduce yourself.

Mike Gioja 1:02
Sure. So Hi, I’m Mike Gioja. And yeah, I’ve been in technology for over 40 years, and I’ve been specializing in HR payroll benefits related software products for probably the last 25 years. I’ve been on both the builder and creative of software products as well as the purchaser. So I first got introduced in HR payroll benefits when I worked for Fidelity. I went to go and work for the CIO on a special project to bring in house payroll and benefits. They wanted to build a business for their clients on the 401k side, that’s when I first got experience with HR payroll. From there I worked for SAP and then I’ve been in a variety of best of breed solutions always in the HR payroll related space and I’ve been here now with Paychex for about the last 11 years and for those that may not know Paychex very well. they started out in the in the payroll business and were founded back in 1971 to provide payroll services for small businesses and since then has grown and evolved to have a Full HCM technology suite of solutions that covers payroll, HR benefits, retirement 401k, insurance agency and so forth. We have about 670,000 customers today over 100 locations and about 16,000 employees

John Sumser 2:14
That’s a big company. So, the heart of the business is payroll, is that right?

Mike Gioja 2:18
Yes it’s been payroll. And I would say we’ve made major investments over the years now to really build out the HR set of products and services.

John Sumser 2:26
So to me, that’s a really interesting idea, because for my money, the most important and interesting data that a company owns is its payroll data. It tells you where people are and what they do and that sort of thing. Can you talk a little bit about how that emphasis on precision that’s part of being a payroll company flows out into the way that you think about the expanded HR product by the paychecks, a discipline that you guys have, that may not be a part of everybody’s view of HR tech?

Mike Gioja 2:55
Oh, yeah, no, that’s a good question. There is a lot of regulatory things that are outside of the business. depicts payrolls. So we’re also in the form of K were in health and insurance. We have a, you know, a very large agency in the top 25, where the number one 401k, record keepers, small plans, all of these also have their own regulatory aspects and compliant aspects and hiring and going through that process, as well as also eo. So there’s a lot of still regulatory compliance, we have a very large organization that helps us with that to make sure at the local, state federal level that we remain compliant in those areas, and, you know, build our products and services, especially from a self service point of view to ensure that employees admin CPAs brokers and so forth have what they need, and at the end of the day, can entrust upon paychecks to maintain the accuracy and the compliancy of the the reporting and the content that they have.

John Sumser 3:51
Now, I am deeply impressed by your work with intelligence tools but it sort of I’m surprised that you’re good at it. Would you talk a little bit of how you think about AI and how you use those technologies in Paychex? I think you do some interesting things. Okay.

Mike Gioja 4:09
So Paychex in the risk and compliance area that group has been involved in predictive modeling and intelligence related engines for quite some time. They have been building models internally for paychecks for us to look across the base in a variety of dynamics, leveraging the data that you were mentioning, along with other data outside of paychecks, economic data and other indicators, geographic and demographic related information to help us determine between that and our clients, what are the right product to sell at the right time, what clients may be at risk. There’s a whole slew of predictive related models for us on upselling and what clients might be ready for it based on a certain set of conditions that we’ve done a lot of learning from, and then we also of course, tap our HR organization for their knowledge and we have a wealth of knowledge from our payroll specialists and service providers that have been working With and answering questions over the years and so the step that we took down the AI path was to move to the our first step was to leverage how do we leverage the data and knowledge that we have and start to provide a higher level of service to our clients. And and when I say clients in this context, its employees and the administration’s and managers of those clients to really understand how we could get them to the next level and to be more self service and self sufficient. And that led us down the path of creating our flex assistant and bring our flex assistant to market which is a bot which is utilizing a variety of technologies underneath it, whether that NLP AI related technologies and so forth to bring that flexus system to market. We combine that with the knowledge that our organization has had over the years. And what do I mean by that is, well, let’s take payroll, for example. And we either there’s always a cycle, you could be erased and there’s other business cycles like open enrollment Where you will, you’ll be doing your moments for health and benefits. So we looked at both general questions as well as key business cycles. And you know, there’s a group that actually sits down with service providers and tap and taps their knowledge and collects their information in a structured way that we can then put into the knowledge base for our Plexus system to utilize and I’ve seen really good progress and results in answering questions real time for for those employees and admins, for our clients on their behalf and have actually seen a reduction in call rates coming in to questions coming into our service organization.

John Sumser 6:36
That’s interesting. So when you think about using AI, it sounds to me like the first concern is increasing the quality of customer service and reducing internal cost of friction, as opposed to the majority of stuff that I see has to operate additional services to customers.

Mike Gioja 6:57
That’s correct. That is our approach. The other thing that we’re doing is we’re doing a fair amount of robotics work. So we have some teams that are building out workflows, automatic workflows, across the systems on the back end for things that the service organization does when behind the scenes. And we’re automating those processes and spending a lot of time to ensure a very high level of accuracy and data integrity. So we’re going slow, and we’re making sure that we’re whatever we’re asking our robotics to do. It’s something that we already do today in a less automated way. And we’re just taking that in automating it and then having clear exceptions to drop out to those people that have the most knowledge, and we’re probably doing three or four robotic automations, a quarter and we’re finding a tremendous uplift to the service organization. And in the offload that work, which is enabling them to be more complicated with the clients, which is really what we’re trying we’re trying to go is let the clients be able to simply get what they need, and then take that to the next level.

John Sumser 7:57
I don’t want to say unusual, but really most the stuff that I see involves people adding things to existing processes. And it seems to me that the approach you’re taking is how to transform the business and increase the value delivered to customers, but not necessarily to add new stuff. Just stop.

Mike Gioja 8:21
That’s right. And we’re trying to do those things that you mentioned at the right time that’s most optimal for the client. So what we’re trying to do is also understand their behavior and collect data through our flex assistant, we know who’s coming in, what’s the question, they’re asking, did they get a good response? Did they like the response? Did they need to transfer to a live agent which they can do 24/7 365 to answer the question that the bot didn’t answer, and actually, when we transfer it to the live agent, we actually have recorded the entire interaction with the bots so that the live agent has that information so that we’re not taking that individual through the same set of questions. Again, and we’re capturing all of this information and learning more about why are they going to a live agent? What are they really doing this helps us fine tune our knowledge base and also starts to give us some behavioral data, which could eventually lead to other products or services. But our goal and learning this is to your point, as we mentioned, to drive a higher level of service, learn about their application interactions and what questions that they have. And from that actually just begin to continually optimize the user experience and the customer experience, which at the end of the day, would continuously increase our retention and drive a stickiness because we’re optimally addressing what they want. Now, will that eventually lead into predictive up? Yes, we do some predictive analytics today. But we take that very slow and careful because we want to make sure that what we’re predicting is highly accurate and gives them a solid base of information and data. We’re beginning to move into that phase, but we’re going very slow and steady and we’ll do a lot of Proving out ourselves before we actually bring it to market because the worst thing I think you can do is bring this too fast. And folks start to make decisions. And they don’t understand the impact of those decisions till later. And then you reflect and go, my wonder if they gave me the right advice or not. So we’re being very careful when we go through that. And we’re leveraging a lot of data and collecting more data along the way to help ensure that accuracy and integrity of that information.

John Sumser 10:25
So it’s a really interesting model. And it sounds to me like you have nearly 700,000 customers, you’ve got a great big pile of data about a certain range of transactions. And then as you get further out into your product lines, the data gets thinner. And it seems to me the approach is start where you’ve got the most data and use that to build out the adjacent data. And when you do that, then you’re talking about analysis and prediction. Do you see a hurdle out as you get past your transactional data is evolving this approach to AI?

Mike Gioja 11:03
Yes, obviously, as you move up, let’s say that maturity ladder and get into more predictive, you still have to make sure that you have a model, what’s the soundness of that model? How does that model evolve? And there’s so many underlying components to those models? are you tracking and understanding as new data is coming in and what could be affecting the model? And so these are lots of questions that we look at. So I do think there are some hurdles. And so we’re moving quickly forward slowly. I would say in that we continue to roll out more and more knowledge base and flex Asst. We’ve now integrated this with our Help Center to give us that next level of knowledge and understanding. So now there is help content and all of the different ways a user can learn whether they’re coming whether they like videos, whether they like to read an article, or do you want to be taken to write were in the app to do it or in fact, you know what, I just want to tell you what to do and do it for me, whether that’s through voice or through or through wearables. And so we’re going to continue to expand in each one of these categories slowly see how the day that evolved in the models. And as with all technologies, there will be hurdle points along the way that will will meet and will break through that hurdle and get through another, another wealth of capabilities that we can provide, meanwhile, continuing to prove out that our current model is indeed both driving efficiencies and interactions in our current environment today, we have something called unified communications. So we have an omni channel capability. We know all of the interactions our client has with us in a consolidated way, whether they come to the phone and talk with folks, do they have email? Have they been texting? Have they been chatting? Have they been going to the bar? Have they been going to the bar and going to the live agent so we can see all those interactions at a client level and individuals in that client? So I have clear data that I can look at now today? Well, I provided a rich set of 401k content has that content and the questions being asked actually reduce the overall interactions and that the channel didn’t change? And we actually see proof to that. And we we don’t know how far that could go, we’ll continue to drive accuracy of information and answer things that we believe we can clearly answer through the bot and that maturity will grow. But we’re going to avoid, you know, getting into behavioral bots and, you know, changing their attitude along the way based, but we may want to pick up attitude differences if you’re frustrated or not. And these are the kinds of things that we will evolve, evolve through as well.

John Sumser 13:25
So are there big questions that your research and development teams are working on? What are they what’s sort of the big horizon for you?

Mike Gioja 13:34
That’s a pretty big topic. There’s a lot there. Again, broadly, we do see AI as a strong mechanism to drive those efficiencies in the HCM world. We have to continue to improve our data and capture all that information that I said that we get along the way and continue to drive that acceleration and go into new areas. So we utilize a couple tools today. We’re now coming out with wearables and the watch, and there’ll be More and more wearables. And we’re also curious to see where the industry is going to go. I don’t know, you might have seen an article that popped up just in the last few days where apple and Amazon and Google are saying, hey, there’s so many things, we need to provide a standard, let the Internet of Things provide a rich standard of how all of these mechanisms going to interact with each other because we don’t want to build solutions specific to a particular vendor. And I think there’s so many things that are exploding in this space, we got to really watch where they are, where they’re going, what’s the standard to see if you’re getting locked in on any particular path on on a particular technology? And the key thing is to make sure you’re abstracting your applications from this to continue your overall growth, if that makes sense.

John Sumser 14:43
It does. So does this mean that you are looking forward to and starting to embrace open standards? Is that what you just said?

Mike Gioja 14:52
Yeah, we certainly would. I think there aren’t a lot of open standards in this space. We have no open source and open standards that we do and contribute to a lot for our applications and our continuous delivery and what we do from a DevOps perspective, and so forth, so we’re heavy into open source and open tools here, we’re really utilizing particular tools through some third parties and waiting to see where open source is really going to go and when are we ready to kind of jump on those and leverage that because certainly it’s a lower cost of development and the technology which we can pass on to our clients as well and prospects.

John Sumser 15:26
So the competition for development talent that does this kind of work it’s insane and I can’t imagine that a sales pitch that begins with, “Do you want to come to work for a payroll company?” is it good lead when you’re competing for development talent? So how do you get people to come to work for you?

Mike Gioja 15:44
That’s a great question especially in today’s you know, economic conditions, it’s very hard to find talent and to pull talent to here and part of it is certain positions you would like to centralize a located and then other positions can be worked from home. And so that’s the first division and we look at our jobs. Do we need them here? Or can they work from home? But we do everything now from? Okay, how can we improve our branding from a talent acquisition perspective? So certainly we work with universities and locals here are where we are located in Rochester with our it and others to, you know, we’ve really changed the branding of a payroll company that’s very service oriented to a company that’s now technology oriented, enabling service. So we do so many things from the leading edge. And the best way to attract talent is the fact that you’re not only dabbling in these areas, you actually have production deliverables, and you’re continuing to evolve those productions deliverable that this is where the place to be is to take the things that you play with and bring them to reality. And we’re doing that in so many areas that does help us acquire talent because we are really building a leveraging this technology with a strong architectural approach where we can grow those technologies across all our different business lines. But it’s it’s tough to get this talent and you know, in other spaces, it’s getting harder and harder. So I think a lot of it has to do with what are the technologies you’re in? Are you competitive? And are you actually going after positions in jobs that are clear and understandable to the industry versus what you may call those individuals internally. So that’s a lot of work. So we continue to work with HR, we work with other departments, we constantly double checking through vendors about our branding, and how we can continue to drive technical talent. And obviously things like these podcasts and other interactions as a company from a technology point of view. Lets people understand that all the here’s a company that really is serious about it, maybe I should take a look at them.

John Sumser 17:33
So you’ve been at this a long time, the technology has come along, I imagine that your thinking about intelligent technologies has evolved over time and is continuing to evolve. Can you talk a little bit about how that evolution has happened for you how you think about this stuff has changed?

Mike Gioja 17:52
That’s a good question. How do I how do I look at the landscape for what you’re doing, what you’re going to do for the businessses point of view?

John Sumser 17:59
How do you look at the landscape and the kinds of things that I’m seeing are, people waking up and going, “Oh, there’s a consequence to this technology that I didn’t quite imagine at the beginning.” Oh, you know, so I see that in your work, right? You have incorporated this technology that most people view as a producer of new and novel value as a way to aggressively streamline and transform your company until there’s a third way of thinking that’s there. And then as we go forward, that we start to see that these tools tend to be the carriers have bias and that it’s very hard to root it out. There may be some things that start to emerge about containing the technology so that they don’t spiral out of control.

Mike Gioja 18:48
Yeah, okay. So very broad but, I get the point. I would say to the latter point, we are concerned about bias. You can see that in different models. That’s why we’re going very slow and careful on anything predictive Whether we’re going to say predict that this candidate is stronger than that candidate, or this one is a stronger reference than that one, or you should hire this individual versus that one, this one’s really ready to be a flight race. So you better do the following actions. We’re being careful, very careful in that way, because models can create bias, and models change and evolve based on data coming in. And so one of the things is, how resilient is it that you understand that model and how things are actually being calculated and derive and have oversight against that? So that is one of those hurdles that we’re not at yet. And we’re waiting to see how technologies evolve in that space. And there’s so much opportunity elsewhere. And when we look at paychecks as a company, when we look at our own service organization, because we’re driving efficiencies and dropping interactions with our client base from one channel and actually handling that through the bot. Let’s say that the real goal of that is working with our service organization saying what are the things that drive you crazy that you spend so much time on administration. natively that we can assist you with because we’re trying to take our service organization to more competative approach with our clients and moving up that ladder from a service organization and changing so but we need to do is we need to move into that slowly. And we need to educate and train our employees along the way about the changes that’s happening to them in that space and give them the training that they need to be more consultative. Many of them have those capabilities, they just don’t get the time to do it. And actually wish they could have more proactive dialogues with the client about Hey, so what’s going on? And so yeah, I understand you’re going to be growing. Tell me a little bit more about that. Let’s spend more time about it, where can I help you and so forth and letting the clients be able to balance the way they want to balance self service on their own and leveraging paychecks as an organization from a service and a competative approach versus we’re going after this particular efficiency to cut this entire organization out. Right. So two very different approaches, and we’re really taking a more conservative approaching a balanced approach into things that are painful for our employees and trying to drive a better work life balance and train them along the way for the next things that we’re evolving to

You’ve raised a number of really interesting questions that we are not going to have nearly enough time to dig into. So let me close and ask you to just cut on the things that you think are the key ethical issues in your work and we’ll draw the conversation to an end.

Sure. So. I would say quickly, move forward, slowly. Make sure you utilize technology so that we’re doing today, whether they’re NLP based AI, AI based RPA based, make sure you have a good solid point of view about what you’re going after, and trying to balance that to understand the implications to you and to your clients. We’re very much driven by what we want for our clients and what we want for our service employees and get your hands into those technologies and start to create production, deliverables. Look at it, architecturally, look at it soundly, and then evolve those in a careful manner. And that’s what we’ll continue to do in integrating our assistant across Help Center and the applications and other things forward and then see where we go from that into the predictive world again, slowly and carefully.

Thanks. Thanks for doing this. Mike, would you take a moment and reintroduce yourself please?

Sure. Hi, I’m Mike Gioja. I’m a senior vice president here at Paychex responsible for all of the data center operations, product development, and security.

Thanks. I really appreciate you taking the time to do this, Mike, and thanks, everybody for listening in today. We will be back here two weeks from now. We’re going to take a little time off for Christmas. Thanks for tuning in. We’ll see you here next time. And thanks, double thanks again Mike. Bye bye now.

Thank you.