In this edition of The Prism, we will begin to explore some of the abilities that will become essential as robotic process automation (RPA) evolves and the digital worker of the future emerges. When we consider how the digital worker will develop, it is helpful for our understanding to try to categorise these abilities somehow and give them “human-like” attributes – at Blue Prism we call these attributes “Digital Skills”.
Don’t think of this as an attempt to view the digital worker as a replacement for humans, but as an enabler for the digital worker to become an intrinsic part of the fabric of a future workplace. The digital worker must at least be able to exhibit some of the same skills as the human workers that it will work alongside.
The first category of digital skills we are going to explore are Knowledge and Insight.
Consider a task that all of us perform on a daily basis – processing emails. How many emails do you open that have a consistent structure and predictable rules that govern how you will process them? The answer is probably very few. For you to determine the appropriate course of action to take, you will almost certainly perform several cognitive tasks without thinking about it:
- Your brain is processing the unstructured data within the email and using years of learnt knowledge. This entails an understanding of your native language, but also very likely some level of “domain specific” knowledge that you have about your industry or job.
- You’re automatically extracting key words, sentences and emotions from the email text, to enable you to gain an understanding of the importance of the email, whether it is positive or negative and what action to take. You will make immediate decisions as to whether this should be something that you prioritise, defer, or send to the junk folder.
- In some cases, you may need to refer to some supporting information to determine an appropriate course of action. Perhaps a report, or some data that helps you understand the history of the request.
So, how does this relate back to the digital worker and skills?
Firstly, it lets us address the problem of language. In simple use cases (for example an email with a structured form) it may be possible to program a simple set of rules to instruct the Robot to continue processing, based on a search for a word or phrase. However, in many cases you will not have this luxury. This is where Natural Language Processing (NLP) comes into play. NLP is an ability that is surprisingly easy to access right now within the modern digital ecosystem.
Blue Prism supports integration with all the major cloud AI platforms – Google, IBM Watson, Microsoft Cognitive Services and others that offer an on-premise solution – such as Expert Systems Cogito. This month we are releasing a comprehensive package of pre-built integrations that will make this easier than ever to achieve — a simple drag and drop exercise. Our integrations give you the ability to easily extract a stream of unstructured text and process it in a number of ways. Primarily the abilities that this supports are:
- Translation – the ability to automatically identify a language and translate it to another;
- Entity Extraction and classification – the ability to locate and classify “entities” or key phrases within a block of text into categories such as names of people, places, organizations and other common categories. The more advanced capabilities also enable you to look for the “intent” within the text. This is where the real value starts to emerge;
- Sentiment Analysis – the ability to identify emotion from a sentence or block of text.
One important point to note is that there are variations between the platforms in terms of their abilities and strengths. For example, it is trivial to create a custom language understanding model using the Microsoft LUIS service. IBM Watson enables you to do something similar with the Watson Knowledge Studio, but this is not yet possible for the Google APIs (but we expect soon will be). Google on the other hand has strengths in the Computer vision space – partially due to the recent advances attributed to the application of AutoML to their underlying image recognition models. In addition, each is instrumented in a different way. Our platform enables you to easily select and mix/match between the capabilities and build these into your processes in a simple and repeatable way.
The quality of this packaged AI is advancing at pace and can deliver significant value.
Customer Example: Our Partner Avanade have successfully used this approach to build a solution for a large insurer to automatically handle email based insurance requests and inquiries. Blue Prism collects the email when received and then uses the Azure Text Analytics integration to perform analysis of the unstructured text.
Based on the analysis, Blue Prism can automatically create a new case in the existing support management tool, with the appropriate details. The customer then receives a response with the case ID. The benefits of the solution enable transformation of the process, with no expensive integration project, add insights into digitised data and an overall improved customer experience. It marks the start of the start of a comprehensive journey to business improvement through intelligent automation.
Let’s explore another example of the importance of Knowledge and Insight in the digitally transformed business world – the ability to understand meaning and context within a specific business language domain. For example, consider the word “Case”. This could have several meanings within different contexts – in an email to a travel agency, “Case” could be indicate subject matter related the loss of a suitcase during a holiday trip. In an email to an insurance company, it could be referring to an insurance case. An important differentiator between RPA tools and cognitive capabilities is this ability to understand context, and this is where use of the standard language processing APIs could also potentially be insufficient – at least not without custom models.
Blue Prism’s partner ecosystem is unparalleled in this space. We are working closely with our AI partners, so that in the future you will be able to extend the standard NLP capabilities to a more specialist model. You can already integrate with Expert Systems Cogito, which can be deployed with a business specific language understanding model.
Customer Example: A large insurer uses Blue Prism and Cogito jointly to streamline their case handling workflow. Using an integrated solution we are able to identify and extract key entities from the unstructured text within multiple input streams (for example, a car registration number, or details of a health problem) and deliver efficiencies to the Case handling process.
Imagine a human worker being able to understand any of the 23 languages being used today by more than half of the World’s population (there are 6909 living languages if you are interested), extract meaning and derive precise emotion from text. Impossible? Certainly not probable at scale.
The Digital Worker of the future will be easily able to understand any language, will be able to extract meaning from text – relevant to the business problem at hand – and will be able to predict the optimal emotional response.
It will be able to reference other distributed systems and bring together contextual information to support handling of any process and present this back to a human colleague as required.
The future digital worker will empower humans and scale your digitally transformed business.