Earlier this month, NICE—which markets RPA under the term ‘Advanced Process Automation’—confirmed that one of its biggest government customers, a major public sector agency in the UK and which it also claims was one of the first significant robot users at scale in the world, had renewed its contract with the firm.
We said, ‘at scale,’ and we’re not joking. This engagement involves collecting and co-ordinating data from no less than 19 different systems across the organisation, and the processing of millions of citizen requests at peak annual periods. No wonder the company describes it as “one of the most complex automation projects for which NICE advanced artificial intelligence (AI) and robotics technology has ever been used”—and which in the last couple of years has become even more crucial to the public sector entity involved, as it now involves the processing of more than 28 million pages of hand-written forms at an amazingly high 99.4% accuracy, as well as providing thousands of contact centre advisors with real-time guidance for Her Majesty’s subjects.
Unfortunately, client confidentiality rules mean we can’t name or give too many details of the organisation NICE is helping here. But such use of AI and unattended automation for eliminating cumbersome, time-consuming tasks, boosting productivity, and improving operations, even if the specific use cases involved are unique to the customer, surely has lessons for other customers.
RockingRobots went to NICE’s Head of Robotics and Process Automation, Oded Karev, to see what could be abstracted out from this on-going project for the wider automation community. The following is an edited version of our conversation.
What’s your role at NICE, Mr Karev?
We used to help agents perform better by automating their mundane tasks and freeing their time to talk on the phone and so give better service. And as the RPA market emerged somewhere around 2015, we decided to carve that work out and have a standalone business, and at the time I was the director of corporate strategy for the company. I was asked by our CEO to lead this new fourth NICE market, that so for the last six years I’ve been the general manager of this division within NICE doing RPA.
You have significant engagement with the UK government, I think?
Yes: we have multiple government agencies working at different scales. Some of them are relatively small or are starting and scaling, and some of them have reached deployment of thousands of robots.
What can you tell us about this contract?
This customer is one of the earliest adopters of RPA globally—they’ve been a customer of ours since late 14 or early 15. Interestingly, their main reason for engagement—surprising as it may sound—was not efficiency, though efficiency is a common reason for customers engaging with NICE RPA; it was, and remains, to improve the customer experience of UK citizens using its services. They want to make sure that their employees have enough capacity to augment and help the public, even in peak times.
Interesting. What does that look like in practice?
This is a customer with thousands of robots, both attended and unattended, and which has launched multiple automations. They have their own COE (centre of excellence), a large population of internal developers. They own the license; we don’t build automation for them—they own it. And literally hundreds of different processes already automated by them, starting very small but which has reached a very large scale.
What’s new, then—what prompted you announcing the relationship?
The first five years of term had come up for renewal, and I said to them, look you’re probably one of the largest, most advanced RPA customers in the world, and you’re very happy: we would like to make use of that!
Ah, very understandable! The handwritten forms aspect is interesting—we believe you are working with a partner?
Yes, Hyperscience is a technology partner of ours which takes optical character recognition to the next level, which we work with on multiple projects. This organisation wanted to extract information from a very large number of written submissions, so we arranged a trial of its scanning tech where it was found to be the most accurate and versatile. In that test, across multiple types of forms with over 400 fields, it delivered reading accuracy just above human accuracy level, but 100% automation of data extraction and export.
Before this software, the organisation was very happy with us but still wanted to cut down on as much error as it could. Our solution was only missing a word every 150 to 200 words, so you aren’t up to 100% accuracy. But once you capture a street address wrong, or you capture a name wrong, the entire automation could fail because there are certain data elements that are mandatory to complete. So, by adding this new module they can both reduce errors but also increase what they can automate. And obviously, a lot more processes can be automated by analysing handwritten forms.
Pretty significant achievements, then. What is the ultimate significance of this engagement for RPA, would you say?
That a very traditional, highly unionised, very large government agency that you might think of a think of as stagnant, very bureaucratic and that all they care about is maintaining the the roles of their employees and the bureaucracy, is very forward-looking and always looking to improve the quality of life of both its employees and the citizens. So, you can still be a large government administrative organisation and be a thought leader and heavy user of RPA.