Improving CX Using AI-Powered Automation

David Worsell: Good afternoon and welcome to Improving Citizen Experience using AI Powered Automation. My name is David Worsell, and I’m gonna be your host here today. By way of an introduction, I’ve worked in govtech now for well over 20 years in a large number of data capture and digital transformation projects. First as a technical solutions architect, working hands-on with a lot of the technology there. I did that for the stationary office for a large number of years. And then more recently, I’ve been the UK managing director for Granicus. I’ve worked across both local and central government, including digital projects for Gov UK in the early days of GDS, HMRC, Parliament, DVLA, DVSA, and loads more, Birmingham City Council and so on.

David Worsell: Over those 20 odd years, I’ve really been amazed at the rate of change in technology, but also in public sector that’s brought about by the change in the technology there. Without wishing to dwell too much, austerity, while it was horrible for those at the coalface, did drive innovation in a sector that was traditionally really slow to adapt to change. With potentially more restructuring to come, the role of technology is more important than ever before. With me today, I have two excellent speakers. I’ve got Gareth Hole from Nice and Nay Odutola from Hyperscience who will be discussing how intelligent automation that uses AI and ML can be used to automate document processing and increase our understanding and provide insight into huge volumes of data that is processed by the public sector each and every day.

David Worsell: In simple terms, intelligent automation is the application of technologies like artificial intelligence (AI) and machine learning (ML) to support robotic process automation, otherwise known as RPA. It effectively makes software robots smarter and it offers greater access to insights and data, whether that data is structured like in IT systems or unstructured data in the form of hand-written paper forms. The concept can be taken further to include process and task mining, which helps us understand where automation might be best applied. It can also include the augmentation of staff by helping them as they work rather than just automating that work in itself. That means you can help staff to achieve things that they couldn’t do before on their own. Intelligent automation goes beyond traditional automation and brings together decision making, decision support, data analytics, understanding of citizen interactions and real time augmentation to deliver transformational business results and an exceptional citizen experience.

David Worsell: Ernst and Young recently published an online survey of about 1300 government and public sector employees from right across the UK to understand their views and attitudes towards automation. Respondents included employees in local government, the civil service, police services, and higher education. Ernst and Young also carried out more in-depth face-to-face interviews with leaders in the civil service, the national health service, and local government. The results of the survey give a really good understanding of where the public sector organizations are in terms of their current route along that automation journey.

David Worsell: Gareth, can you tell us a little bit about yourself and your role at Nice?

Gareth Hole: Thanks David, it’s a pleasure to be here with you today. I’ve been involved with business transformation and enterprise software for over 20 years now. The last eight of those years has been with Nice, where I focus specifically on innovative uses of robotics and AI to really help organizations deliver the best possible services to their end customers. A lot of that work has actually been with the UK public sector and the likes of HMRC, DWP, Defra, Police Forces, you name it, to really help them deliver a better citizen experience.

David Worsell: And Nay, can you tell us a little bit about yourself and your role at Hyperscience?

Nay Odutola: Thanks Dave. I’m the general manager of Hyperscience for our EMEA business. I’m responsible for our go to market strategy and helping large organizations from Global 2000s to government agencies across Europe. Hyperscience is an intelligent automation company headquartered in New York with offices around the world. We enable Global 2000 organizations and government agencies to automate mission critical processes. Working with state of the art AI and human in the loop technology, we classify, extract machine readable data across diverse documents.

David Worsell: Technology is developed to solve problems. In the Ernst and Young survey, respondents were asked about their most important challenges facing the public sector over the next three years. As we can see, they ranked operating within their budget first, meeting increasing demand for services, and improving the way they serve citizens, partners, and communities. Gareth, does this come as a surprise to you? Why do you think these are the top issues for the UK public sector?

Gareth Hole: No surprise at all. With budgets under pressure departments and agencies must do more with the same or less, even as the demands on services are increasing. We’re also seeing that the advances in automation come at a time when globalization, demographic shifts, geopolitical, environmental and pandemic threats are all really challenging the landscape for the government and the public sector. These drivers are clashing with the more immediate fiscal challenges that they have. If you look at the keywords used in parliamentary debates over the last 10 years, resilience has now actually overtaken efficiency in how often it’s mentioned. That’s not too surprising given the challenges that the public sector faces now with the pandemic and also supply chain issues. That means there’s a realization that whatever is done in public sector needs to be adaptable and agile enough to cope with future challenges.

David Worsell: And Nay, what about citizen experience? What are you seeing when speaking to your public sector customers?

Nay Odutola: Citizens now expect the same personalized service from the public sector that they’re getting from the private sector. If you order something today online, you can either get it the same day or the next day. There is not this notion of waiting around. There is a need to optimize public sector staff time and efficiency by automating manual processes and also extracting better quality data across multiple types of documents such as handwritten forms. The demand has also significantly increased by the pandemic as well as Brexit. Using myself as an example relating to passports and driver licenses, just to make a name change on your driving license is still very much a manual process where it’s a handwritten form that you need to complete and send off. There is over a 10 to 12 week backlog just in terms of getting that document back from the public sector.

Nay Odutola: Public sector organizations are expected to process millions of forms such as application forms, tax documentations, and many more. These documents come in a wide variety of formats. They’re often messy, handwritten, and have lower resolution. This could be a scan of a scan, making it really hard to read, extract and process that data. The data remains very much untapped from a resource perspective, and many public sector organizations are struggling with this. The result is a really strained system, overworked employees and frustrated citizens waiting for answers.

David Worsell: We’ve looked at the challenges being faced. I guess we should look at the benefits now of using intelligent automation. The Ernst and Young survey highlighted IA applications could potentially deliver benefits in several areas. Nay, how does automation deliver on these benefits for you?

Nay Odutola: The public sector processes are drowning in paper and a lot of information. A lot of these processes are manual today. To efficiently digitize records and keep up with demand, there is a need for intelligent technology that can read diverse and complex documents, which then translates into structured data that’s ready for processing and analysis. Intelligent automation solutions like Hyperscience help automate these diverse and complex processes by classifying, extracting, validating, and enriching information from incoming or even archive data as well as handwritten and machine printed documents. Government agencies are increasingly turning to intelligent automation to breeze through data, gain efficiency, improve citizen and employee satisfaction, but also to drive better outcomes.

David Worsell: Gareth, if you’ve been involved in a number of public sector projects, where do you think solutions like Nice and Hyperscience have the potential to help the most?

Gareth Hole: I think there are three main areas that are of particular interest right now for the public sector. The first is AI driven solutions that are quick, agile and deliver real world benefits. For example, by combining Hyperscience with Nice robotic automation in a recent public sector project, we were able to read over 400 handwritten fields on scanned forms. Each of those forms was over 16 pages each, some up to 200 pages. We’re able to do that with a 99.4% accuracy, which is unprecedented. We’re also able to extract 100% of that data and perform the necessary work across multiple line of duty systems automatically. This level of success in reading unstructured data we haven’t really seen before. People actually only typically read with something like a 96% accuracy. So we now have technology doing things that people can’t on an ongoing basis.

Gareth Hole: Secondly, it’s about helping staff as they work and not just automating things. We’ve seen a significant increase in demand for having robots help people as they work—augmentation rather than automation. At Nice we have NEVA, the Nice Employee Virtual Attendant. Software is very good at churning through lots of data and guiding people on complex rules and regulations in real time. We’ve got 1700 NEVAs helping in another government department with appeals cases for benefits. They have a “bundle builder” use case that gathers documents for a caseworker to work through an appeal. The result is half the time taken to complete a case, meaning the caseworker can focus on the important parts. It’s also a better audit trail.

Gareth Hole: Thirdly, it’s about identifying where to automate and where to help staff. This needs to be unbiased and based on data. We’ve developed machine learning technology that can identify sequences of work tasks and processes that happen in an organization. We can automatically prioritize these based on real world data: how often they actually happen, how long they take, how many systems are involved, what type of work it is. You then have all of these different opportunities bubble to the surface in that prioritization algorithm. When they do, you can click a button to automate them or provide NEVA to help someone to complete those processes.

David Worsell: For me this is absolutely fascinating. Going back into the dim and distant past, I was working on a project for the Commonwealth War Graves Commission scanning casualty records from the first and second World War. These are historic documents, sometimes nearly a hundred years old, written on very poor quality paper, lots of handwritten stuff with blood stains and mud. At the time it was kind of an all or nothing. You could use OCR to scan some of the documents, but they were just error strewn. What we ended up doing was going back and rekeying absolutely everything manually. It became a huge complex exercise with teams of people, really slow and still full of errors. This is just a great example of where technology now is bridging that gap between full automation but including humans at the most essential part of the process to make sure the quality and the accuracy is right. I just wish I’d had this kind of technology way back then.

Gareth Hole: Just before you move on, I was reading the other day that the Church of England is actually going through a similar exercise. They’ve got something like 16,000 graveyards and they want to digitize all the information in those parish records as well, which I think is gonna take them something like seven years. Nay maybe one for you guys to have a think about as well. There might be something you can do to bring the seven years in.

Nay Odutola: Absolutely. That definitely we want to look out.

David Worsell: Nay, I’ll ask you that question as well. Where have you seen intelligent automation help the most in the work that you are doing?

Nay Odutola: Gareth summarized a lot of those points nicely. I might just focus on one area which is identifying and selecting which of the processes to automate. There’s so many different processes. If we look within the public sector remit, it is trying to understand the art of the possible. Traditionally support functions such as finance, human resources and procurement have been the primary candidates, but there are also major benefits in automating aspects for areas around service delivery and operations too. When looking at intelligent automation, you can streamline a multitude of processes including passports, IDs, driver licenses, court documentations, disability benefits applications, individual and corporate tax documentations, assistance or loan applications as well as healthcare claims. Especially now with the pandemic, areas like disability benefits applications are gonna be key and there’s gonna be a lot of influx and high volumes of data.

David Worsell: I guess the reality of where we are now is kind of different. Public sector organizations are now at very different stages of their automation maturity. The Ernst and Young report shows that the good news is that lots of organizations, about 34% of them, are already implementing some level of automation. Alarmingly, 22% still have no plans whatsoever. Currently within the public sector, more and more organizations are looking to utilize automation, but so far very few have managed to deliver this at scale. Obviously there’s a whole load of obstacles to adoption. Gareth, why do you think public sector organizations struggle and are they any different from the private sector organizations that you’re working with?

Gareth Hole: In the public sector, there’s sometimes a reluctance to embrace or embark on large scale transformation programs mainly due to past experience with big old monolithic IT systems that can take years or even decades to implement. There are parts of the public sector who have almost resigned themselves to the fact that technology enabled change is hard and slow and expensive, but that doesn’t have to be the case. What we do with Nice robotics doesn’t require any changes to existing systems at all. It isn’t dependent on significant IT effort. That means you can use it closer to the operational areas and it can be changed quickly to adapt to changing rules and regulations. That kind of agility gives excellent resilience.

Gareth Hole: A lot of what we do is based on robots helping people as they work. There is some concern about automation if it is purely automation trying to replace people. But interestingly enough, I think there are actually more areas being concerned about the results for the citizen. How can a robot do a better job than the person with years or decades of experience? It’s a perfectly valid question and a fair challenge to traditional robotic process automation solutions. People are great at doing some things, like having a conversation and understanding nuances. But robots are great at gathering an awful lot of data, analyzing it, providing guidance on rules and regulations. A combination of automation for mundane work and the likes of NEVA as a virtual helper for the workforce is actually a very positive one. When people have their own personal robot helping them to achieve their own KPIs, the penny drops. It’s less scary and quite a positive experience.

David Worsell: It is easy to level at government organizations that they’re a little bit slow and a little bit behind, but I think the reality is they’re actually doing some really innovative things out there. Some organizations like HMRC are at the cutting edge of quite a lot of the stuff that’s going on there. Given the large scale that they operate at and the risk inherent with that, the fact that they’re doing some of these really groundbreaking things is absolutely fantastic. Nay, what do you think is stopping public sector organizations from unlocking this value in automation?

Nay Odutola: I wouldn’t put it down to one thing. I think there’s lots of different variances stopping it a bit. Sometimes there’s a lack of education. If we track back into what Gareth talked about in terms of machines doing some things really well and humans doing some things really well, it is sometimes maybe the notion of people thinking “am I gonna be replaced?” I think it’s more the education around actually it’s complimenting what I’m doing. You’re not going to walk a 200 mile journey, you’re gonna use a car which is a robot effectively. It helps you do that, but it still needs your help. That’s the way people should start to look at intelligent automation—humans and robots working together in harmony to get a better end result and be a lot more efficient.

Nay Odutola: Public sector organizations are faced with large amounts of variable documents and they’re messy. These come from handwritten forms requiring lots of different types of automation, and standard automations today like an OCR or an RPA can’t do it alone. If you look at the amount of data that are actually readable, it’s something around 20% within the public sector. That’s why Nice and Hyperscience have partnered to offer an end-to-end automation solution that enables organizations to scale their automation initiatives.

David Worsell: The Ernst and Young proposed five best practices to unlocking the potential of intelligent automation, and these fall into two main areas: strategy and people. Getting both of these elements right is gonna be absolutely essential to delivery of a successful project. Nay, what are your thoughts on this?

Nay Odutola: Strategy is always key. Having a clear strategy and implementation plan are some of the key fundamentals needed to translate the vision into reality. If you don’t have a clear cut strategy, most of those projects are gonna fail. A strategy should set out the scope and purpose of why you’re looking to implement a technology like this and the potential value that you’re gonna deliver. It should define the organizational structure to develop and help you scale the use of automation across your enterprise. Look at the bigger picture: your overall vision to deliver onto your core mission and your operational goals. What is your current problem statement? What is the impact? What are the required capabilities you need to address this? What does good look like and how are you gonna measure success?

David Worsell: Gareth, you mentioned earlier on the importance of people in the process. What are your views on that and how do we make sure that they’re included in the process and taken forward with it?

Gareth Hole: One of the most important activities in automation projects is identifying and selecting which processes to automate. Traditionally you look at the people doing work in finance and HR and procurement as they have a lot of mundane work. But there are major potential benefits in applying capabilities to aspects of service delivery and operations as well. With no disrespect to EY or other large consultancies, I don’t think there’s a need to pay huge amounts of money to identify the right processes. The traditional organization wide review can now be done using technology. With Nice analytics desktop analytics, we can generate data automatically based on how people are using their existing systems. We can bring that together automatically to identify the right processes to focus on. This is done very much in a bottom-up approach based on looking at the people and how they’re interacting with systems.

Gareth Hole: As processes and regulations change, the analytics changes and what bubbles to the surface is new each time. Even as you get a changing organization, you can apply the same approach. All you need is about 20 people doing their work using whatever systems they’ve got. Within a couple of weeks there’s enough data generated to uncover some really great opportunities. I always recommend having operational staff involved in the project from the very beginning. You need that human perspective on what that means to the end citizen and the people doing the work. They understand the challenges, the systems, the view from a citizen perspective. They should form part of that change process and it really should be something that they’re actively engaged with, not something that happens to them.

David Worsell: Introducing disruptive technology like intelligent automation can have a major impact on an organization. In this sense, it’s less about the technology and more about the adoption and the cultural change needed to support it. Would you agree with that, Nay?

Nay Odutola: You just hit the nail on the head. One of the most significant parts in every organization is culture shift. To adopt a new technology requires a shift in culture and mindset. This will ultimately lie with the departments that will have the responsibility for executing the strategy. By being transparent and adapting working practices, employees will feel more confident and they’ll understand that the technology is there to work for them rather than to replace them. The lack of education forces people to think “am I being replaced?” rather than “the machine is being brought in to help me improve my capability and efficiency.” By having a culture shift, they’re more likely to start pitching ideas for automation. The first step is to make sure your employees understand the art of the possible so that they can relate the technology to their own pain points that they’re facing on a day-to-day basis.

Gareth Hole: Having a clear ownership structure has a lot of benefits. Any technology does need some governance around it. The real benefits are around creating reusable capabilities. For example, if you want to automate a lot of handwritten forms using Nice Hyperscience, don’t treat them individually one by one as separate projects. Treat it almost like building a machine once that you can then feed any new forms into. If you then measure the impact and the benefits and the accuracy and the reduced turnaround times, that also builds trust. Once again, you can see the machine is working. There’s no substitute to having a robot on your own desktop. That’s a great trigger for people understanding what it can do. Having a robot on your own desktop doing the work alongside you means that when you make a promise to a customer, you always keep it because that robot has done the work there.

David Worsell: We’ve spoken a lot about central government projects and large scale projects, but is this something that can be applied at a local level, Gareth?

Gareth Hole: Absolutely. Local government faces a lot of the same challenges plus some which are very specific to the services that they provide. They’re also very close to the end customer, the citizen that uses their services. That means they need to find opportunities to automate a lot of different areas. That can be done using the analytics we talked about before so they don’t have to incur a lot of these large consultancy bills. With things like click-to-automate, they can also reduce development costs and time to deliver. I think the advances in technology have actually lowered the barriers to entry to using this, and that’s a great thing for local governments.

David Worsell: Moving on to questions. First question I’ll ask to Nay: who owns this function? Who owns automation technology? Is it IT or someone else?

Nay Odutola: It typically depends in an organization. If we look historically, some people go to IT, some people go to the business. In the day and age that we are in, IT have their own internal customers as well, which is typically the business ’cause they’re gonna be the end users of this. So it’s kind of a two prong attack. It would be owned by both departments: the IT side from an architecture and implementation perspective, but also the business in terms of the day-to-day executing and providing that end citizen and customer experience.

David Worsell: Gareth, you mentioned earlier on about analytics and how you could find things to automate. How would you get started with that?

Gareth Hole: It is actually quite straightforward. If you’ve got let’s say 20 or so people doing the same kind of work, first talk to them and tell them what you want to do. After that, it’s just a case of installing a small piece of software on the desktop. It doesn’t interfere or change the way they work at all at that stage. It just gathers information about how the systems are used. You let it run for two or three weeks, all the data is gathered automatically, and then effectively you look in the portal and there’s a prioritized list of these opportunities to see what’s being discovered. Then you can generate the automations or the augmentations to show the people involved how they could be helped. End-to-end, typically it’s something like a four to six week exercise with minimal IT effort.

David Worsell: Really quickly in one sentence, what’s one piece of advice that you can give based on your IA experience?

Gareth Hole: Don’t assume you know exactly what intelligent automation can do for your organization. If you make assumptions like that, you could limit what you can achieve. We are now achieving things with robotics and AI that simply weren’t possible even 12 months ago. So talk to the people that know the latest capabilities and have used them in the real world, and be open-minded to the fact that not all RPA, not all robotics, not all AI capabilities are the same.

Nay Odutola: The latest innovations in artificial intelligence and machine learning have the ability to bridge the gap for success where legacy tech typically fails. So definitely talk to people that know this and have implemented this into the real world because that’s where you’re gonna get the best advice from. Don’t just think you know all the capability on AI because it’s constantly improving. There’s always a new capability being released. So definitely do some research and talk to people who are using it on a day-to-day basis.

David Worsell: Technology is really just a means to an end. What’s really important is how the technology and people interact. As a result, the cultural changes are absolutely critical and maybe more important than the strategy and the technology that drives them. Gareth made that point earlier that it’s augmentation rather than a replacement of people. The new breed of automation tools aren’t going to completely replace humans in the way that we thought they would maybe 10 or more years ago. But what they are gonna do is make them smarter and faster and a lot more efficient. This will allow people to focus their time and efforts on the things that really matter and that’s delivering great public services. 10 years ago it was all about OCR and there were loads of error rates. Now we’re pulling people into that process and getting it right. These new systems using artificial intelligence and machine learning to intelligently process and understand the data, even handwritten data, are smart enough to know when that person needs to be pulled into the process. It’s smart augmentation. The benefits of this approach are significantly higher quality data that can be used to weigh decision making—faster, smarter decision making. This delivers better outcomes for the organization, but far more importantly, better outcomes for the citizens that they serve.