Experiences with Robotic Process Automation - from Getting started to Keeping going
By Jonathan Kidd, Head of digital operations & robotics, Bank of Ireland
Jonathan Kidd, Head of digital operations & robotics, Bank of Ireland
In the context of business process automation, AI is being used to refer to everything from Chatbots and Voicebots to Natural Language Processing & Generation (NLP & NLG) to Machine Learning and yes, RPA. I find it useful to group these components into three broad categories:
1. Human Interface
Technology which supports the interaction with the person. Topical examples are chatbots and voicebots. We need NLP, NLG and a variety or other tools to facilitate the interaction.
2. Business Process Fulfilment
Getting the job done! RPA is the silver bullet in this space right now. This category also includes machine learning so that the dumb robots ‘learn’ to do better over time.
3. Data & Learning
This is the part that is closest to ‘real AI’. Using machine learning and neural nets to identify patterns upon which future decisions can be made. This includes teaching the fulfilment robots what to do. But it is more about identifying patterns to make better decisions in everything from consumer purchasing to complaints peaks to IT outages and anything else where data can be captured and analysed and responses then improved.
RPA means using a software ‘robot’ to complete some or all of a business process that are person is already doing. Often the robot just does the the repeatable, boring bit and leaves the person to do the smart bit – we use RPA to take the robot out of the person.RPA is like an excel macro except with RPA, the macro can work across multiple systems and isn’t embedded in any one application.
Benefits of RPA
We now categorise RPA benefits under very traditional benefit categories:
1. Cost Saving
RPA does save human effort but industry experience indicates that this is often savings through cost avoidance, growth absorption and offshore savings rather than material reductions in core headcount.
2. Customer Experience
Faster response times, fewer errors, consistent responses all lead to happier customers provided we use it for the right things. Customer doesn’t want to discuss a complaint with a robot but they probably don’t mind if the backend complaint logging is done by a robot while the call centre agent talks to them or to another customer.
3. Risk Reduction
Robots don’t make mistakes. If you give them the right instructions, they’ll get it right every time. And if they come across something they don’t recognise, they’ll tell you they’re stuck and ask for more instructions.
4. Staff Development
RPA (and the other aspects of AI) opens up new career opportunities for staff.
Robots are involved in some or all of over 100 business processes across the enterprise and we continue to finding ways to use RPA as we better understand the capabilities and as the tools themselves improve
Getting Started
Bank of Ireland is ‘relatively mature’ in RPA terms. We have been involved with RPA for over three years. Robots are involved in some or all of over 100 business processes across the enterprise and we continue to finding ways to use RPA as we better understand the capabilities and as the tools themselves improve. We now face a variety of challenges which I touch on below. But in terms of getting started I would include the following as some of the factors that help us get successfully off the ground:
1. Senior Executive Support
Similar to any emerging technology, we needed the time and space to test and learn. This meant a [very small] ring-fenced team, elevated access to systems and access to people and processes which were good candidates for automation.
2. Outside Help
We supplemented the internal team with one or two real experts in the field to helpget us off the ground. There are lots of stumbling blocks and pitfalls in getting started with RPA. Many of these relate to how RPA interacts with the rest of your operating systems and applications. One or two people who have been doing this for 3-5 years or more savedus a lot of time and helped get our people up to speed much faster. If you pick the right partner they will also help you put standards in place that will allow you to scale once you havehad the first few successes.
3. Hero Process
Basically, make sure your first process makes a splash. Find just one process that meets the criteria below and make it work. Prove that RPA isn’t just a curiosity but that it can make a real difference.
Finding the Right Process
So how do you find the right process? We try to follow these principles in identifying candidates for automation:
1. Structured Digital Input
The process has to have a digital source. This could be a web form, SharePoint form, email or even as OCR’d PDF. The more structured the source the better the conversion rate. A strongly validated web / mobile form would be a great start.
2. Significant Manual Effort
Notwithstanding my comments above about RPA not being just a cost play, it is where everyone starts. So make sure the process you pick has enough volume and requires enough human effort that it will have a big impact when the work is being done by the robot.
3. Existing Process
Pick something that people are already doing! An existing process is working today and the exception scenarios are known (and hopefully documented). An SME will be able to give a full set of instructions to the robot (via a developer) instead of just the ‘happy path’ or something new. Unearthing the exception scenarios for new processes is very time consuming.
4. Rules-based
The process has to be rule-based, requiring little or no judgment. The robot doesn’t ‘think’. It does what we tell it to do. If we can’t explain it by ‘yes or no’, we are going to have a problem.
We have found that the more processes we have automated, the more these principles have been challenged. RPA becomes the solution to every problem and expectations mount. Other factors come into play. But the above are great guidelines to get started.
Keeping It Going and Where to Next?
RPA is now part of our Operation al Excellence toolkit. It sits alongside BPO, Digitisation, LEAN, Centralisation and workforce management. We face a different set of challenges and questions as we become embedded across the enterprise. Some of the areas of focus today are:
- Benefit Tracking– Continue to prove the business case
- Governance – IT or Business Architecture?
- Operational Risk Framework and Management
- Innovation – where next with RPA and how to replicate this success for other AI tools
- Resource Training & Retention – in a world where RPA remains a growing industry
For me, the most pressing topics centre on Operational Risk Management and Benefits Tracking & Delivery. There is still a lot to go for in ‘just RPA’. But the most interesting are probably the relationship with IT and the evolution into AI. Ultimately RPA is one more tool in the enterprise toolkit which we leverage to serve the customer and deliver sustainable profits for the enterprise in a safe manner.
