On-Demand: Identifying the Value of Intelligent Automation Opportunities in Your Company

It's very rules-based. It's very structured. I call it the swivel chair where you're sitting there with three screens and you go from screen to screen performing different functions. Essentially, a software bot is think of it as a digital worker. It is essentially emulates what someone does today, a task that you perform, so that essentially you can have the bot execute things like it can run Excel, it could run a macro, it could send an email, it could post an entry into a general ledger system.

It could provide notifications, it could perform workflow. There are a number of functions and tasks that it can perform just like a human worker could today. The real value in that is that, automation allows us to improve scale, also to enhance our controls. Bots as we know, wouldn't take a nap, wouldn't rest, won't can take a break, go and get a cup of coffee. It's constantly on, it's constantly working and they don't make mistakes unless the program itself was in our ... So the reality is once it's stood up, it'll function continuously and that's powerful. Some other areas in terms of deploying this technology, if you just think about things like data management, all of you probably have different systems, you have your accounting system and then you have your operational systems.

Maybe you have a sales system and emails and Outlook and things like that. Time all that together, all those data points, moving information across systems and bringing it together, that's another powerful way of leveraging these bots. We'll talk a little bit more about data in a bit, but this is a powerful tool to help us manage data better, enhanced controls and efficiencies. What I would say is that, it is geared towards rules-based structured data processes that if you were to sit down and write them out on a whiteboard and you could list out the 10 steps and show me how to do it, then I can essentially go in using a technology like a UiPath or Blue Prism, which are two RPA providers. These technologies can be used by the business to build these process, automate these processes and deploy these. This is definitely something that can be leveraged by the business.
Jason Juliano: I'm going to go into, what is AI and putting AI to work. In terms of artificial intelligence today, you have to think about as imitating some type of human behavior, allowing the machine to predict certain things as a human will and providing the ability to perform a specific task that is intelligent or smart. How it actually does that is through several different methods. It could do that through machine learning where you basically populate data into the machine learning model and the system will understand attributes of that data, whether it's structured, unstructured, and then figure out specific context within the data that it's reading. Then from there, you can do deeper learning where you could have like AI solutions teaching other AI solutions.
The other method is a natural language processing. Many of you have already experienced that with Alexa, Google Assistant, your voice recognition within your car, speech to text, text to speech type of applications, chat bots. When today when you're calling a medical center for instance, now a robot will answer it and it'll utilize artificial intelligence to gain what you're trying to say in terms of that intent. That's natural language processing.

Then finally, image recognition where it's reading things like pictures, forms, documents, similar to, using Google Photos today where now Google provides you the functionality that it understands who the person is within that picture. It can scan thousands and millions of photos in literally seconds. Like humans, AI systems are not perfect. They need to learn by observing, adapting to their environment. The data, again, typically looking at unstructured data, structured data. Unstructured data could be anywhere from contracts, purchase orders, invoices, pictures, videos, audio files, policies, procedures, regulations.

Right now, we feed in our solution with massive amounts of different regulatory standards from, anywhere from FINRA to HIPAA, HITECH, NIST standards, ISO standards, COSO standards. Again, it's like training a toddler and you have to provide it the right data. If you train a toddler with the wrong information, the toddler is just going to have bad habits and similar to really training the system, AI system to provide better findings. There's two methods of training, you could do supervised training or unsupervised. Supervised training is basically having a domain expert work with the machine learning model to help it learn.

Then unsupervised is just giving the system a massive amount of data sets and having a system come up with their own findings at first. The value that you get from that is deeper insights into your customers, deeper insights into your business, reduce risks, potentially reduce fines for your firm, increase client and employee satisfaction, empowering your business to make better decisions today. In a recent MIT Sloan survey, they found that many businesses across the board from small businesses to enterprise companies, 90% of decision makers within those businesses are saying that they need a digital strategy and AI strategy in place to be more competitive.

Greg Fritsky: Yes. Thanks Jason. I spent a lot of time this past year speaking at a number of events and then speaking with participants that are very much focused on this whole space of AI and intelligent automation. One of the top questions that people come up with is, how do I identify the right use cases, right? How do I understand where I can deploy this type of technology and get some real return on investment? Oftentimes, you'll hear people say, "Well, we're kind of going through our own transformation, we're looking to enhance, improve our processes." The question will come up, "Should I automate after I go through that deep dive and transform those processes or should I automate first or do I do something and do them in conjunction with one another?" I often say that I think it's best to look at this as a transformation exercise where you can look at a process, how it is today.

Because at the end of the day, our processes were developed because we always did them that way. They'll oftentimes just sit there and say, "Well, we're not even quite sure how we got to where we are and maybe there's reasons because of legacy applications or we didn't have access to certain data or certain systems that maybe that automation allows us to do that and do it more effectively." One of the things that you want to consider too is, where are people spending their time? The biggest concern, initial concern I hear is, "What's this going to do in terms of people in their jobs and their functions?" The reality is, it's going to free up time. It's going to release capacity. There may be tasks that people do today that they're obviously not going to do anymore, but now maybe they can do additional control-type functions.

Maybe they could focus more on the data. Having an understanding of what the outcome is and what value your business is or what the value of this type of program is going to provide, it's very important to start with, because you don't want to go down the path of trying something, realizing that or getting everybody excited and then you want to move forward and then you have to stop the program short because we're not sure what the new operations model is going to look like. There's a number of things to consider in terms of defining your journey. When you look at the processes themselves and you evaluate a process, go through that whiteboard exercise, identify those 10, 20 steps, take it as far in 10 as you can and understand which process steps are specifically just doing and which ones are actually thinking.

If you can identify, 70, 80, 90% of a process that's really just doing and you can automate that and then that 10%, well, maybe that 10% stays as a human function as it should be and you can enhance that. Maybe there's additional data points now that you can extract and you can use and leverage. Maybe there's additional controls you can put in place. Maybe it's a process that you did on a monthly basis that now you can do it more frequently, maybe more, maybe on a weekly or even daily basis. There's a number of things to think about as you enhance that process. Then ultimately you want to automate as much of it as possible, so that you've essentially transformed that process. Then you'll hear organizations that might have different geographic centers that are looking ... They'll say, "We do it a little bit differently." Everybody will say that, but if you come from an automation template and you standardize that as much as possible, it's hard to really deviate from that.

If there are deviations, they're localized and they're pretty small and they can be fitted into the model. You can program for those deviations and you can handle that. If you can automate, again, 80, 90% of that process and use that as a template going forward, the efficiencies gained are tremendous. It's really, again, looking at where people are spending their time, understanding which processes are going to provide you the most ... If they were to be automated, provide you the most return on investment and potentially which processes are at risk if we don't do anything and we don't change them today or automate them. There are number of things to consider.

One exercise that I like to work with clients is to go through an exercise of where do we start? What would be that phase one activities. The easiest way to look at it is, where people spending their time, where there are a lot of FTEs, their time being spent, but also where the processes are less complex. What I mean by complex is that, out of the gate you want to have that initial project, that initial success to be achieved within weeks, not months, not years. Because the reality is, this is a very agile type technology where you can find savings very quickly by just automating 60, 70, 80% of a process in a matter of weeks and then showing the organization what you've achieved and now you can go after maybe additional steps in that process or expand it to other areas.

It's all about release of capacity. Then you expand the program to other that taking inordinate amount of time, but maybe they are a little bit more complex, meaning it may take a little bit longer to implement that technology. The fact is, what you learn in the first phase is something that as your organization develops its knowledge base, and maybe you will even bring in or create a center of expertise, you can get engaged in this and actually be part of the buildout. This actually helps with the cost aspects of developing this technology over time. Then you look at the other processes as well. If you tackle those first five, 10 processes, you're going to get a lot of attraction and you're going to see folks get pretty excited within your organization. That's the exercise that you want to take.
Jason Juliano: All right, so where to start. Finding Intelligent Automation Opportunities, and it's really looking at taking these RPA type of tasks and defined processes and integrating them with artificial intelligence. I mean, to start with as Greg stated, you really have to understand that your current business processes, internal and external, and they have to be well-defined. Potentially, before you even begin with looking at RPA, you have to really focus on your business and understand your internal business processes, looking at tasks that are mundane that you can apply it to a robotics process automation bot. Then figuring out what piece of that can't be provided through a process automation bot and what can be leveraged through AI. You'll never replace your knowledge worker, but you're augmenting their skill set with utilizing tools like process automation, artificial intelligence.

If you start this off correctly and really spend the time to defining your processes and looking at those mundane tasks that can be automated and looking at the pieces that can be ingestible through AI machine learning models, you'll really get a good return on investment on that, you'll gain trust internally and externally with clients. You'll build up confidence within your practice areas and then you'll really create a digital strategy foundation for your firm. If you look at the charts from left to right, when you start off, you have all these manual tasks and you have to really well define those manual tasks, what they are. Now once defined that as I stated to before, you take some of those mundane tasks and you create unattended robots that will create specific tasks. It could be perhaps doing data entry in an Excel Spreadsheet, reconciling some geo inputs.

Then from there, you look at those processes from beginning to start. That can't really be fully done by a process automation tool that will need attended robots to do the full tasks. That could be a compliment of looking at emails, perhaps put doing some data entry into a CRM solution like Salesforce. Then taking that small piece and figuring out, okay, we have this unstructured data and structured data from our internal process, off-road process and we could have the machine, not only train a bot but will also provide us deeper insights into the data that we're capturing, providing more value for our business.
Greg Fritsky: Yeah. Just to add to that, I think that's ... It is very valuable to look at how you can bring a cognitive type solution to this discussion. Oftentimes, we see client ... You focus on the first phase of, "Let's get the tasks tackled" Right. Then we've achieved that. That's pretty significant. Then it's, "Okay, what else can I do now with the data that I'm extracting? How can I take this further? How can I help resolve some of the exceptions? How can I look at things like risks or tolerances or other insights?" That's really where you start bringing in the cognitive capabilities as well. That's the whole intelligence in this conversation. It's important to understand that these are steps and stages, but always thinking in terms of, "How much of that overall process can I automate and where do I bring in some of those cognitive augmented intelligence-type activities?"

Just to mention real quick, so we here at the firm are using a technology called UiPath. It's one of the leaders in this space. There's Blue Prism, there's automation anywhere, there's a number of technologies and they're all very good and they all accomplish similar results. We're using UiPath and we've actually had a number of case studies that we've done and use cases we've done for clients as well, which we focused on again very repetitive manual tasks that took substantial amount of time. Typically, we will look at things that would equate to the ... Equivalent of one FTE performing that function. Again, when you're looking at your use cases, if you can achieve the savings of one ... The equivalent of one FTE in a year, you're going to get back a very positive ROI.

That's typically where companies will start. In this case, it was a Dunning Letters. This particular use case was just generating Dunning Letters. It was a very manual intensive tasks that some individual or individuals were performing. You just go through the process looking at ... Using a legacy application. There are a lot of different touch points and sending out thousands of emails and just the coordination of itself, just that workflow and just monitoring what's coming in and out, which was an absolute nightmare for this particular customer. Looking at RPA, this was an obvious choice, so we went through the, what we call a Process Mining Discovery Workshop where we just went through and we said, "This was an obvious choice." By developing in just a matter of, honestly weeks, we were able to deploy a bot that could essentially perform those activities that the individuals were.
That also allowed for a number of reasons it saved. Not just saves the time of people spent on the activity, but the legacy application they were using was, they were able to phase that out, able to automate the sending out of all the letters. I think this is an important point, increasing the frequency of how often they can perform that function and tasks. Oftentimes things fall behind and we lose out because we have things like age receivables or bills that need to go about or invoices to follow up on. Then just the sheer volume we can't handle. Allowing them to be able to do this more frequently and more efficiently, was definitely a cost savings. That's one particular use case.

Another use case, we're going to talk about a couple of dimensions of this, but Revenue Recognition and I'm going to talk to from an audit standpoint.

Here at the firm, we also have ... Just like you all, we have a number of very repetitive manual type internal audit tests and audit testing that we perform. We are looking for ways to deploy intelligent automation to help us to drive value for our customers, so that not only are we performing the doing, but we're deriving the value of what the data is telling us. Also, when you think about just performing things like internal audit tests in this particular case, pulling together, extracting data for sampling. In this particular case, it was a client with a number of branches that has to perform this exercise pretty regularly and every time they do so, it takes countless hours to just to pull together the data and the samples as part of their internal tests by perform ... By building a bot, it essentially performs it on behalf of the client, can extract the data, can actually test the total population and perform that and sheer like minutes or as opposed to days or even weeks to perform something like this and real cost savings.

Something that was estimated savings, something like 500 hours a year just for this one particular test. This is just automating the test function. We're actually looking to expand that to bring in some of the intelligence to actually perform more of the ... Doing the query into what are the anomalies, what are they ... What is this telling us? What are the potential risks applying against a baseline or a benchmark and performing some of the additional audit testing that we perform as part of this review? These are results that we're doing in collaboration with a client, but it benefits both the firm and the customer as well. Jason is going to talk a little bit about what we're doing with regards to 606 and revenue recognition reporting here at the firm.
JJ: Yeah. For EisnerAmper, we created a machine learning models to read in revenue recognition FASB guidelines. There were a lot of changes down the pipe from the ICPA and we are taking these guidelines to really understand the contracts related to revenue recognition. As I stated before, it's unstructured data. These contracts come in via fax, they come in via scan copies. They have handwritten texts on them. Some machine with Watson will understand the context around these elements within the contract and then relate them back to a FASB guidelines.

Well, EisnerAmper's journey with utilizing our artificial intelligence specifically IBM Watson, we created several engagement opportunities with a basic contract analysis, revenue recognition, lease accounting, employee benefits and all these practice areas. Again, relate back to specific FASB guidelines where we taught the system how to look at this from an accounting perspective. As the system reads these contracts, policies, SOC reports, the audit logs, it understands the context of what that language is and We're providing the system a full deeper insight into the full engagement, allowing the auditor not only to get deeper insights but to get consistent artificial intelligence findings and help them augment that specific engagement through that audit ... On practice area.

All right, so we're going to go into the AI journey. That one? Yep. Okay. Greg and I are both, we're going to talk on this, but as we talk to clients from a day to day perspective, we really get them to understand a process of bringing on AI into their firm and augmenting intelligence and what that means. We look at their existing solutions, whether they're enterprise resource planning solutions, a CRM that they have in place, understanding internal and external data.

One of my pet peeves as ... I tell them, we look at this from a data first strategy because if you don't understand the data that you bring in and you basically use bad or inconsistent data to train the machine learning models, bad data is going to be coming out. As I stated before, it's like teaching a toddler, you teach a toddler bad habits and it's just going to consistently provide you a bad experience throughout that process from a process improvement standpoint. Greg, we'll go into this, but yeah, it's really to first understand and mapping out your full process end to end. Not only from the enterprise but from a line of business perspective.
GF: Yeah. I think, I go through this journey map if you will with folks. Oftentimes, we hear a lot of good things about what these technologies can do, but you need to understand, you have to go through the blocking and tackling first before you go out for the touchdown pass. Right? Oftentimes, I have people say, well this is great but we're not comfortable right now with the current systems or the data that we're getting out of the system or the processes themselves or we feel people are doing a lot of tasks that we don't feel that they should be doing. How do I get my arms around all of this? We oftentimes, if we just historically speaking, we look back at, maybe you've put in an ERP system like an SAP or a net suite or PeopleSoft over the years
We've invested so much money in technology. People have a little pause when they think about, okay, where am I going to make an investment? I want to make sure it's ... I'm a little risk adverse here in terms of jumping in before it's well baked. We have our ERP systems, we have our applications, but it's ... I often say, don't even worry about the technology to start with the process, focus on the process itself, how you do it today. Going through a discovery exercise, going through what I call process mining, of taking that whiteboard, going to the whiteboard and working with your team and just developing an intend view of what do we do today and how do we do it? Then looking at ways to eliminate the tasks that are just, again, the doing and how do I enhance that task by bringing some level of data or intelligence to it and how can I achieve that.

Once I get that, I do this current state, Future State exercise. We use the Future State as a blueprint for, okay, you ... Knowing what we know now and where we want to go. What type of technology do I need? Right? Do I want to look at an enhancement to what I already have? Do I want to look at a replacement strategy? I see this a lot of times, we've spent money, and we have great technology. They don't talk well to each other. The data's hard to pull together. I don't have one single source of the truth. Managing data is probably one of the top issues I've heard with regards to clients' challenges. Before you even get the data and analytics, it's like, how do I manage the data? How do I access it? How do I harmonize it? How do I model it?

Big data, cloud mobility, the acceptance of these technologies. You've probably started looking at. Possibly you're using one today, a software, as a service, the cloud model. I only use what I need. This is gaining more and more attraction. It was, there was a time that no one could ever envision putting financial data out on the cloud. Now that's become a reality. Or a more cost effective way of deploying a technology and using what you really need as opposed to studying up all these new systems and applications and then all this data that ... Pulling all that together and managing that.

Where are we today? Well, again, we have the robotics now, the machine learning capabilities. We're starting to introduce these things and these things, again, putting this roadmap together of how do I achieve this type of efficiency is really about making sure I lay out a good ground ... The ground rules of how do I manage data, how do I transform these processes, and ultimately where do we want to go organizationally? What's going to happen to the people that are doing these tasks? What roles are they going to play? Oftentimes we see process improvement teams spin up from that.
JJ: Greg, I'm going to take you back to the second tier, which is one thing that's identified here. Business process management and really just you're going over understanding and ongoing changes in your processes and managing that because another use case ... Yeah. We see a business process change over the life cycle and let's say we created a bot for this. Now the business process changed but the bot is still doing what it's doing. That's the opportunity where you, only have to understand your business process, but you have to align it to the technology solutions that you roll out to help you provide a more value. Again, there's multiple touchpoints today. Yeah. Everyone's on a cloud. People have multi-cloud environments, people are utilizing their mobile devices for like Salesforce data from a big data perspective. There's data coming all over the place.

It's coming internally, externally, and people don't know where to start at. Again, we tell our clients to look at a data first strategy, but it's really to understand these separate data silos from a master data management perspective and then figuring out, all right, you have this data warehouse and what does that mean from internal, external perspective? Do you have to create a full data Lake? Maybe you have to take chunks of this database with this database to provide deeper value to the business. From the dirt tier, that's when we go into a cognitive solutions, artificial intelligence, creating a RPA bots for these business processes, utilizing natural language processing, natural language understanding to understand the context of the data that the systems are reading. We've recently created a module called the artificial intelligence multimedia enhancer.

The specific use case for that was to provide law firms a way to upload audio and video files of expert witnesses in court cases. The system will provide them, what's the intent of what the expert witnesses saying, whether it's an insurance case or a malpractice suit. Really understanding those facts and circumstances within that case and then allowing the attorney more information to provide them a win for the case. Then we go into artificial intelligence and how that's augmenting the workforce with augmented intelligence where you have, again, AI models, teaching other AI models. You want anything to the intelligent automation piece.
Greg Fritsky: I would just that one of the things that I'm hearing more and more at conferences speaking with people is less about artificial intelligence and more about what I would call augmented intelligence, which is really understanding that this technology is not replacing us. It's enhancing what we do. It's enhancing how we think. We're going to continue to provide a value to these processes and these services as humans. The reality is the technology is such and can crunch so much data and provide us with so much value that it's essentially allows us to now bring to our clients additional value, additional insights, predictive modeling type capabilities. These things are constantly changing. Think of it in terms of how it augments the way you work as opposed to worrying about it replacing you. I think that's a mindset. We talked a little bit about this. We're just.... Okay.
Jason Juliano: This slide basically provides you a deeper insights into contract examples. These are examples of a generic contracts that were uploaded specifically to understand the elements in an intense, again, the context around what the document is saying, anywhere from who the parties are, the obligations, exclusions, what rights do each party has, what right does the landlord have, what right does the tenant have? Then deeper insights into audit clauses, intellectual property. We're doing this across different factors of document types that include anywhere from leases, contracts, master service agreements, employee benefits, regulations, policies, procedures, even SOC reports that we were able to read and now I have the machine provide deeper insights that saves you time. Let's say that auditor attorney's reading in a contract and it takes them like a full four to six hours to do the full analysis of that contract. While with augmented intelligence, the machine is providing them findings on some of those consistent elements and contexts of these documents. Then instead of spending four to six hours on this, they're doing the full assessment in under an hour.

All right, so this is a specific another use case that we have, and Greg mentioned this previously in some use cases and utilizing a robotics process automation for revenue recognition, reading invoices, but then this takes it to a next step further, creating intelligent automation process. Then we do that with specific things like creating RPA bot that will handle email automation or error handling, doing discrepancy reporting, image recognition with computer vision, creating a bot that understands what it's reading on the screen, assisting either account rep, a loan processor or underwriter and helping enhance that work or to be a digital worker with augmented, a digital assistant that work in their behalf.

This specific use case is not just creating a process automation but looking at machine learning on model that reads in invoices, a machine learning model that reason of purchase orders next line. Okay. These are examples of how are we doing that today with example, invoices and example, purchase orders. Again is taken these fax, scan copies of invoices and purchase orders and porting it into someone's ERP solution to further and do a full analysis on that just the specific invoices and purchase orders. Then finding the discrepancies next line.

This slide shows you on specific reporting on where the machine with in conjunction with the price of automation bots. Whereas able to find discrepancies and some of the use cases could be vendors submitting additional invoices. Then what the PO was approved for. In this example, the system is telling the net account rep that it found six attributes that do not match and a need further assessment, this specific wine was an invoice for a higher amount that flag that wasn't approved via that purchase order that was signed off on.
Greg Fritsky: One of the questions that we've been getting is, in terms of your teams and skillsets, given that ... Depending on what your role is, if you have an accounting finance team, how do you staff, what is the particular skill sets need it for this. What I would say is that, and I advocate as is the center of expertise development of somebody within your organization with a passion for technology that's probably from the business that would drive this, help develop those technologies, basically the liaison between the business and those developing the bot technology. It could be somebody from your team, it could be somebody from technology, but I would advocate it's a consortium of different individuals.

You may even want to get somebody involved from the controls and risk side of the house if you're looking to enhance controls and that kind of thing. I would also say that since data is a big component of this, we're hearing more and more of this as folks that are responsible for data and being your data stewards absolutely should be part of this exercise because this is really going to allow ... It's just all about data. This is all about how to manage it and how to process it and ultimately how to use it to get cognitive capabilities and better insights. Those individuals that are responsible for the data management, putting this type of technology and should be part of that organization as well.
Jason Juliano: Yeah. There's no lack in regulatory standards there for our massive file today and it's just increasing over time. Right. As Greg mentioned, it is crucial that you really do a deep dive into a data protection, data privacy. The States are mandating that now. California and New York, last couple of months and years and started focusing that GDPR, there's huge fines associated with GDPR and it's not just a specific to Europe, it's anyone that does business with countries, whether they're clients or even a business to business agreement. Going back to following a specific governance model, it's key to, once you engage in either process automation or you move and mature into intelligent automation practice, you really have to work with your IT partners, the business, and the data scientists. Everyone has to collaborate on an ongoing process to make sure, again, that you're getting value from these tools that are getting implemented. In addition, especially from an AI perspective, you're not teaching the systems bad behavior or providing it a bad data.
Greg Fritsky: At this point we'll take some questions. All right. We do have one question. It is, what if our organization does not have data experts?
Jason Juliano: That's where you work with partners, like an Aponia data and even EisnerAmper where you looked at what your business process is and then we'll look at your business process and then we'll help you identify the data elements that will provide you value, but also puts you in a path that will really understand the type. It's an art, right? We have to understand now when your business, but we're in a data's coming from internally and externally and what's the problem that you're trying to solve and what's a value to the business. It was important not only to understand that business, that process internally, but to also understand the data that you want to capture or you are capturing and then work with a trusted partner.

Again, it could be a firm of ours. We're a premier IBM partner. We're also a multi-cloud. We use other solution stacks. EisnerAmper has a lot of practice areas. Even for us. For instance, EisnerAmper is not just a client of ours, but they are a true partner where we work behalf with specific practice areas where there's manufacturing, pharmaceutical and we bring in these knowledge expertise workers to help us compliment what the end solution will be.
Greg Fritsky: Okay. I think we had another question. How do most companies get started with a program like this?

Yeah. I'll take that. The question here is, what ultimately, is the outcome that you're looking to achieve here. Oftentimes, I say, you're not going to know that right out of the gate, but you are going to know that automating a process is going to save money and time and obviously everybody's interested in that. I would say that it's important again, to identify that individual or individuals in your group that are going to be drive this initiative. This is to me, an exercise of enablement. What I mean by that is that this isn't a technology that you should be dependent on consultants. As much as Jason and I like working with our clients, it's more about education, getting you comfortable with the type of technology that's available and then allowing you to implement it a change going forward.

It can be part of a change initiative. Maybe you have an established group that's responsible for that, or maybe it's just something that's kind of siloed in your business and you're looking to just make some enhancements. Either way, what I would say is that you're going to need to identify somebody in a partner or somebody internally that can help guide this program. Certainly treat it like a program. I wouldn't say go in and unlike other big transformation type programs where you put a three year roadmap together, this is, "Hey, in the next three to four weeks, what can we automate?" It can be that simple, but you should simple ... You should also be thinking about how, if this is successful and it will be, where do we go from there?

How do we deploy this type of technology across business silos? Because if we're having here and someone else is using a different technology or a different approach, that's where things become difficult. There's a level of change management. Project management certainly can start this within the business itself, but you definitely want to be looking at it from a project governance standpoint as well. Center of expertise is important and working with professionals, folks who've done these types of programs and built these technologies and make sure that that partner is enabling you and teaching you and getting you comfortable with this type of technology.
Jason Juliano: Right? You definitely want to make sure that this compliments your overall business strategy in the next year or two. You do want to look at that low hanging fruit type of project engagements, but you want to make sure that gets incorporated into your overall business strategy and rolls up. Again, we work with a lot of mid-market. Everyone thinks that investing AI are super expensive, there's things that we've done today to help you get started right away at a low cost to entry. Yeah. Sometimes there's an investment, but sometimes, yeah, it's a small investment.
Greg Fritsky: One thing just to add to that, and this is ... Jason is correct. This is not a big company technology. It's being deployed at all different levels. I would say too that RPA is not a silver bullet solution. There are other ways to automate your processes. There's other technologies. You may be sitting on an application right now that has that capabilities inherent in it. I'm taking a fresh look at some of these type of technologies as you upgrade your current software, maybe looking at how they get the software to work better with other software's in house. Some of that interfacing, data mining, there's a number of different aspects of what automation really is and it really starts with some doing some level of assessment and getting a roadmap. It's defined for you so that you're not jumping all in with the technology but make sure you have a well-orchestrated plan and roadmap. That's what I would suggest.
Greg Fritsky: Okay. Well, we thank you for your time. Again, it's an exciting topic. Feel free to reach out to myself or Jason. We would be happy to talk to you more and provide you with data points on this topic and otherwise please have a happy holiday.
Jason Juliano: Yeah. Thank for your time. Again, happy holidays and if you have any questions, feel free to contact us. I believe our contact information is up. Again, thanks and have a great day.
Moderator: All right, we hope you enjoy today's webinar. Please look out for a follow-up email with a link to the survey and presentation. As our speakers mentioned, if you have any additional questions about today's topic that you would like address, please feel free to email our speakers directly. Thank you for joining our webinar today.

^{Transcribed by Rev.com}