#37 The Intelligent Process Automation Episode

Hello and welcome back to another episode of the Process and Automation Podcast. As most of you know, to all the listeners of our podcast, is that we cover all sorts of different areas of automation, and today we would look, we would like to look at the intelligent process automation. And so what is intelligent process automation? That's what we cover. And why does it matter? Arno, welcome, welcome to the show as well. And why does it matter?

Well, I guess it is a buzzword that's out there. What we need to do first is, is try to look at what is intelligent process automation, because we've heard the term out there, we've heard the term RPA, which is robotic process automation. And, you know, today we, we're going to sort of differentiate between those two things. So intelligent process automation, or IPA as it refers to, is, you know, it refers to tasks that are automated or optimized in part by artificial intelligence and machine learning algorithms. You know, intelligent process automation can reduce human intervention in a variety of business processes out there. And, you know, IPA solutions go far beyond simple rule-based tasks. For example, IPA tools can apply artificial intelligence to process unstructured data, and this is, this is something that many RPA tools cannot do. Or, you know, to provision IT resources to ensure critical SLAs are maintained, for example. Another example can include the use of machine learning algorithms that enable intelligent process automation tool to improve task performance over time, for example. So really it's taking automation and applying cognitive behaviors to it to make it intelligent. So it's not just rule, rules-based, predetermined, but really it works on sort of machine learning algorithms where you train, train it to do things to and, and learn by itself. And in this instance, you know, one of the examples I've mentioned is unstructured data like an invoice, for example, and invoices could be different depending on, you know, your customer, extracting that structured data from the invoice, which has got the line items, the invoice number, doing that for a variety of invoices for the view to give that to a robotic process automation bot that needs structured information to do some processing, right?

Yeah, indeed. So some of us, um, yeah, and some of these processes, and when we look at them, yeah, just RPA is not efficient enough, isn't it? So the bot needs eyes, and these eyes are nowadays— it is one capability of artificial intelligence, isn't it? The vision capability of AI. And so that's a very, very common one when we look at intelligent process automation and how we can enhance RPA. Yeah, really lots of people are sort of using these terms a bit like the same, isn't it? So they say intelligent process automation is RPA. But yeah, as you mentioned, it's just a combination of it. Where—

Yeah, exactly. And I think, you know, intelligent process automation is often regarded, as you say, as being the same as robotic process automation. And it's sort of just half true because robotic process automation is a key component of intelligent process automation. But intelligent process automation doesn't necessarily have to include RPA, you know. So sort of with intelligent process automation, you might just have an AI model that sits there that's not dependent, provide its output to RPA bot. Indeed, RPA bots can, can exist without, you know, the intelligent part, but of course the data needs to be be very structured. And, you know, robotic process automation really refers to the tools, applications, platforms, and all the scripts that automate simple rule-based repetitive tasks. And these tasks are generally quite, you know, time-consuming when you do it manually, you know. So for instance, instead of collecting phone numbers from applications, RPA tool can be can be trained to automate that task, right? But the inherent problem with RPA is the fact that it is still rules-based. And if you update your, say, your contact form, or the customer enters information in the wrong row on the form, you know, the RPA tool won't be able to successfully complete that task because of course it doesn't have that intelligent capability. And this is exactly where robotic or, you know, intelligent process automation is frequently used, you know, at that point where RPA is no longer effective. Because it leverages artificial intelligence, you know, these intelligent process automation tools can complete more complicated processes that incorporates, you know, a variety of new and emerging technologies.

So yeah, let's look at the artificial intelligence and also maybe machine learning and how that can be used with Intelligent Process Automation. Yeah, so the use of AI and ML, machine learning, in Process Automation enables the Intelligent Process Automation platforms to go, as you mentioned earlier, further than just front office and back office tasks. It goes further than just RPA used to automate these, these tasks. Artificial intelligence, for example, makes it possible for, for these platforms to analyze this unstructured data, as you mentioned. So the semi-structured data and the unstructured data are necessary for language, natural language processing. So in detecting the intention, the intent, and other cognitive technologies. So this, this combined, this allows users to build really complex workflows for— and now the topic comes up again for us again more and more— so the chatbots or responding to customer requests. So as soon as we understand what it is the bot needs to do, it could be not just text, for example, it could be also images, isn't it? Understanding video clips, all sorts of use cases are available to to give the bot a little bit more information to process more in this end-to-end process. So where you usually use RPA tools to automate tasks that already exist, intelligent process automation, so together with AI tools, these tools give users the opportunity to sort of rethink the capabilities of existing processes. Yeah. Or to really look at the process and to optimize them with, with with additional learning. So deep learning is the phrase here, or use some new technologies such as intelligent decision-making to create really innovative processes. And yeah, with those capabilities in your platform, when you start using machine learning algorithms that also looks at all the historic data and the real-time data, you can optimize these processes, your process even further. So when we look at this machine learning and looking at real-time data and historical data, we touch, overlap slightly probably with here and there with process mining. But this really allows us to optimize processes in real time and then predict what might happen in the future. So very, very powerful when you think about predictive analytics and that kind of stuff. For example, you could look at your workflow and automatically route the work on the predicted runtimes and all the content you have discovered, all the understanding you have historically. You can manage how the whole process flows to solve problems. And yeah, so this is where you can improve your processes with artificial intelligence and machine learning. Yeah, so that's something you should really look at.

Yes, and of course, I think we have spoken about you know, how important data is. A lot of people say data is new gold. It feels like they've been saying that for years, but I think it's so much more important now to look at your data because of course, for end-to-end orchestration, you really need those data sets across your business. And of course, you know, in today's business environment, These platforms we use, these SaaS platforms, they don't exist as a single island. And, you know, if we deploy an intelligent process automation tool, you know, it will have to access and manipulate data from a variety of sources. And, you know, a lot of this access rely on disparate technologies, you know, from different vendors. So, you know, if you look at intelligent automation and sort of having access to this really valuable backend data, a level of extensibility is required, you know, on the backend, you know, either through your sort of your workflow automation that exposes data, your enterprise job scheduler that does ETL jobs, for example, service orchestration. We talked about data APIs and data platforms quite extensively in the past. So these solutions provide sort of these kind of universal— so most of them provides these universal connectors and API accessibility to seamlessly manage data and dependencies between disparate systems, you know, with inside your your business. And that truly then makes end-to-end process automation possible. And that then unlocks, you know, the ability for you to use intelligent process automation tools, you know, to find that right data that's on your infrastructure that you could, you know, deploy these automations, you know, that you require. And provide and unlock that value intelligent process automation can provide you as a business. So again, it's that kind of back to that thinking where we have to have our data in a good place. We need to expose it in a way that it can be consumed easily. Of course, there are these different silos that exist inside business. It can sometimes be a bit scary when you look at the different disparate data sources, but there are options, you know. Almost all of the modern SaaS applications provide connectors, even with each other. Obviously, legacy systems, there's ways to connect that to your data sources. And if you can't, of course you can use RPA to do that. And again, you know, once you've got this sort of, this data liberated, very powerful place then to, to overlay your intelligent automation tool, where let's look at chatbot, for example. A chatbot can actually index your backend data. We've seen examples where A chatbot understands data structures, it understands, let's say, your customer entity. So it knows how to bring that data into a conversation with, with somebody. And you know, that, that, that's a fairly, fairly basic example, but that, that can get to the single source of truth very easily just by using the chatbot connectors, for example, to connect to your Yeah, so then let's look at a few other examples.

Or maybe before we jump into the examples, how good would all that stuff be if it's possible on a low-code platform, isn't it? So we discussed that lots of times, and sometimes our listeners might think artificial intelligence, machine learning, end-to-end process automation, orchestration, all this stuff might, might be very complicated. But it can be also done on very, very powerful local platforms. Everything is part of that platform, the RPA, the AI, and lots of stuff. So it doesn't need to be difficult. Everything can be done with the power of low-code these days. So, yeah, so then let's look at a few examples here. You mentioned this chatbot example already, a good use case. So if you look at into a few more use cases, these IPA tools or intelligent process automation tools are used to automate, as you mentioned earlier on, or time-consuming and routine business processes, enabling all the employees to really look at the more valuable tasks in their departments and spend more time on those instead of feeling like a robot. And by freeing up this time from your employees, your business, it will be more efficient, will be more productive, and save a lot on your full-time employee costs. So this is a good advantage. Not everyone is necessarily looking at saving saving full-time employees. That's in most cases actually the opposite. They never want to get people— want to get rid of people. They just want to extend sort of the real valuable capacity of people. So that's why it's important. And yeah, so there are lots of different use cases in lots of different industries and lots of different departments. So it could be finance, finance industries, healthcare, manufacturing. So we covered so many use cases as well in the past on on different angles, on chatbots, on low-code, on RPA. So yeah, feel free to just go back on our episodes and catch up on all this good content we have produced over the— yeah, have produced in the past. So, but let's pick one here first. We really like always financial services. So it's a very busy, busy and innovative industry. Sometimes it feels like they are a bit old-fashioned, they're not moving fast, but they actually have the pockets to try all this new stuff. And when they look at it, they come up with really good use cases. So customer support, for example, those professionals in these departments have to gather customer data from databases. They have to retrieve that information from— they have to do phone calls, emails, and online chats. So this is very time-consuming and can impact sort of the whole customer journey, the customer experience, some topic we're really passionate about as well. And this tool can be an intelligent process automation tool, can be used to pull data from automatically, data from different databases and updates these records with all the information gathered from emails and phone calls automatically. So it can sort of transcribe and classify and identify intent and maybe the mood of the person calling. So to store that extra information in a database and yeah, so you can gain so much more from your data.

Yeah, exactly. Yeah, I think financial services is still a very good place to find use cases because of course there's— it's transaction rich. And these transactions tend to be high volume, low value. But like you say, you know, identifying intent in a transaction, for example, from different sources, extracting structured data from that out into a database and then presenting that to a person with an index of some description, a risk index or a mood index. That's very, that's very valuable. And, you know, we see it, for example, in insurance where the claims department might spend, you know, hundreds of hours a year entering data from claims forms into, into the insurance management system or the CRM system or backend system. And, you know, deploying an intelligent process automation tool can be used to scrape all of that necessary data from the forms and automatically port them onto your insurance system or your CRM system. And again, this task can be included in a larger end-to-end process that delivers that relevant information directly to the customer or the end user, whoever is looking at the claim. I mean, I saw, saw an example of an implementation of an AI-driven low-code platform where the, the telephone call was automatically transcribed and the intent extracted from the telephone call and the Attendant Bot actually navigated to the right place to go and update the information. All the agent had to do is really just verify that was correct, and so I didn't even have to type the information in. So it's really, really powerful sort of cognitive services that's coming out and that's available as part of these you know, intelligent process automation platform.

And you mentioned, so the bot is identifying the intent and then doing all that stuff, and then the human workers verifying it. So this is the kind of deep learning which at some point happens, isn't it, or occurs. So, so the learning flows back into the system, or can flow back into the system. So then at some point the bot is achieving like 90% of, or 99, on whatever, success rate. Then at some point you don't need that verification from from the human worker anymore because you then start trusting that algorithm based on that learning, the historic learning. So you can—

Exactly.

It's even more efficient.

Exactly the point, you see. And this is the thing that differentiated from normal robotic process automation is that the machine learning algorithms in the background does constantly learn. So they do learn that for these specific phrases, the intent was extracted. Maybe a human actually confirmed that that was the correct intent. And if that happens again and again, a couple of times, because of course the data was extracted, nobody changed it on the form. So it means that the algorithm was correct. So that just ensures that the learning of that sort of sort of the accuracy level elevates up to the point where, you know, you don't need to have somebody that needs to review it because it's got enough sort of knowledge and intelligence based on previous learnings that that is the intent of a customer that phones in that wants to change the address, for example.

And that's exceeding, and very many often, or many times actually, the accuracy you have from your human workers in these departments. Because at some point, this is what we see in here, at some point, yeah, if you do this day in, day out, at some point you get blind for something. So you make probably more mistakes than the bot would do. It doesn't mean that we as a human are obsolete, but I think for some tasks it definitely makes sense to get it done by a bot and AI.

Yeah, exactly. So let's throw another example in there. Let's say in, in shipping, for example, where you could use an intelligent process automation tool to analyze shipping data, for example, to optimize your shipping routes and schedules in order to reduce bottlenecks and prevent delays and, you know, optimize valuable resources. Now you can, you could, you can, you could automate that process or try and automate it without using machine learning. But of course, you have to look at all of their variations, and it's quite difficult and time-consuming if it's rule-based. Where something like this, if you do deploy the deep learning and you can actually create a model that would simulate how do we optimize assets, how do we optimize the routes, that's when the, the sort of the benefits comes in place. So, so, so 3 very simple use cases. Again, you know, financial services, insurance, and shipping. So quite, quite, you know, different types of scenarios where this could be beneficial. But, you know, hopefully it's clear to our listeners that there are massive benefits of deploying intelligent process automation. I guess the kind of the main benefit I would see is, you know, efficiency, optimization, and innovation. And I think that's generally true for process automation as a whole, you know, where you automate these routine processes that you save employees' time and make their lives easier. And but when you introduce kind of machine learning algorithms, you know, that these algorithms can discover new ways to optimize your your processes for even further efficiency and productivity gains. And, you know, when you've got these capabilities with inside your business, these intelligent process automation capabilities and the tools, you know, both your sort of your business and your IT users that can use those technologies to develop even more innovative solutions and more importantly to improve those customer experiences that we're always talking about.

Yeah, these benefits are quite tangible if you put this in numbers. So you can automate like 50% of your manual tasks or more. 50% is just sort of on the lower end. What we have seen in some areas, it's really going up to 90%, which is fantastic. So you the processing times reduced by 50% and ROI achieved is over 100%. So this is usually in very short time frames. And so I guess that's quite appealing if you look at this. You still need to find out what processes are worthy to improve and to automate with intelligent process automation. But if you can just pick a couple of those to get started in your business, I think, yeah, you want to do more. And then hopefully then the pipeline fills up with lots of processes to automate.

And yeah, so very, very good, good benefits there and good reason to look at And yeah, yeah, like, like we said earlier in our Clubhouse session, you know, we, we see the gap between IT and the business is closing, you know, where IT is deploying more, more of these low-code automation tools, including intelligent process automation tools, robotic process automation to low-code, low-code automation tools. Where, you know, it does accelerate digital transformation initiatives because it is easier to deploy these solutions. And the end of it is, it is reducing costs across the business and increasing some satisfaction. So, so this alignment, you know, where IT is kind of aligning very closely with, with the business to meet their, their needs and to orchestrate these, these end-to-end processes for the users and to streamline processes across the business. It's a really sort of exciting future, seeing those things happening more and more. And I think with intelligent automation in the mix, that will go to the next level where We've got our processes automated now. We've got our robots in the background that can do, do things, rule-based activities very fast. They can work 24/7. This kind of just brings a new level to our business process orchestration where it's really intelligent. And, you know, by combining all of these, these sort of automation tools with machine learning, applications, you know, we really kind of getting to that sort of hyper automation space, where it's very clever things that can be done very quickly.

Yeah, the big thing to look at is really the end-to-end automation, isn't it? So we had lots of point automation within the business, and I think looking at the end-to-end, so that's sort of the really, really big one to look at. And these are where the really huge benefits are, where it gets really interesting to have this, this big mix of capabilities.

Yeah, absolutely. So very exciting, lots to talk about. We'll try our best to, to try and cover all of these different disciplines. But, you know, a very, very, very good topic, intelligent process automation. And I think as AI becomes more accessible to people, you know, this would become or will become, you know, I would say very mainstream very soon, you know, when people realize its benefits and, you know, how it how it can sort of take automation to the next level, right?

So yeah, we recommend to our listeners to catch up on the previous episodes. If you're not a regular listener, follow us on theautomationguys.net to just be informed of what's going on in the space with events and content we share, webinars, our podcast episodes. Everything you will find there. And as always, if you need sort of help and want some advice on how to get started, Arno and I, we are here for you to go on a session and just show you what is possible, thinking outside of the box. Those kind of sessions. And yeah, feel free to reach out to us at any time, and let's automate it. Unfortunately, that's it again with this episode of the Process in Automation Podcast. If you like this episode, please give us a 5-star rating and don't forget to subscribe to this podcast so you don't miss any upcoming episode. We hope I hope you will tune in next time, and until then, let's automate it!