Taking the 'dumb' out of Robotic Process Automation (RPA)
Robo Advisors 962 X 514

Taking the 'dumb' out of Robotic Process Automation (RPA)

18 November 2019

Robots are dumb. Fact. Don't believe me? Don't fret however. The good news is, that Robots are only the start of the story.

Take a look at our previous article if you haven't already to see the thinking behind that statement. 

The tip of the artificial intelligence iceberg if you will.  What's really exciting about RPA, is how it can dovetail with the other buzz technologies of today, namely; machine learning and natural language processing (NLP).

In order to try and justify my reckless usage of arguably the two biggest technical trends of today in a single sentence, let me explain how Equiniti ICS is working to augment RPA with these technologies, using complaints as an example.

Complaints, up until five or six years ago used to be relatively simple for organisations to manage, following a tried and tested methodology.  Clients or citizens would be dis-satisfied with an element of service, seek to contact said organisation and provide a 'strongly worded letter' to that effect.  The organisation would have a defined process (or alternatively as I've witnessed, a rather panicked email based escalation chain!) in order to respond to the complaint and carry out investigation if required.  As a technical solution provider, we could genuinely help organisations by providing portals on multiple devices to allow the complainant to submit their complaints, backed up by a governed workflow, allowing the organisation to demonstrate how they managed the complaint, carried out route cause analysis and addressed any underlying deficiency within their organisation.

So far, so 2013.  Today, I invite you browse the Facebook and Twitter hashtags of organisations and central government departments, not to mention the rise of TrustPilot.  It turns out, that we as customers, like the rest of the world to know about the dis-service we're being subjected to.  I put forward the case, that as a citizen, I'm perhaps not as interested in receiving an email back from an organisation assuring me that they are addressing my grievance (honest),  and more interested in letting the wider social circle know of my ordeal.  It’s certainly much easier to post my strongly worded letter on Facebook... I post stuff there all the time, so it’s comfortably in my skillset.

So how do we, as organisations, pivot to manage this?  What about the investment we've made in complaint management systems?  This is what Equiniti ICS is actively looking to address through the latest technology.

We've already seen how RPA is excellent at providing cost effective and cheap interfaces, without having to invest in complex API's. That's why, we're using it proactively to retrieve the complaints that you don't even know about. i.e. The Facebook posts tagging your organisation and the TrusPilot review with a one star rating. The power of this capability, is only matched by the simplicity behind it. Remember, RPA automates tasks that we (humans) can carry out. If I can find those complaints on social media, then RPA can. The difference is that RPA can comfortably cope with the volumes and bring them back to our complaints platform, where we can now manage them.

So that's us up to date.  2019.  Great.  We're sorted.  Hope you enjoyed the article...

Sorry... Not quite!  If we just take our complaints from social media and subject them to our complaints process, we're still basing our handling of complaints on a very traditional methodology.  Remember, RPA used standalone, isn't actually intelligent.  What we need, is something to help us make decisions better and continuously improve. 

That's why Equiniti ICS is actively working on using Natural Language Processing (NLP) on the complaints that RPA is identifying.  Complaints are an ideal use case for NLP.  As a society, we don't tend to write simple one line complaints, especially when we know our friends and the wider social circle will be reading them.  We'll add detail, colour and most importantly of all emotion, or sentiment.  The NLP model will read not just the 'text', but the 'language'.  From this, it’s looking to determine a couple of key items to drive an automated, intelligence led process. 

First of all, NLP looks to identify multiple complaint types.  We're doing this to ensure that a complaint doesn't just get submitted and managed by a workflow process that either isn't addressing the underlying complaint type, or doesn't recognise that the customer actually has multiple complaints within their statement.  How often have we as customers, felt 'passed around' an organisation, corresponding with individuals who tell us that we need to be passed to a different department..... for the third time that day.  NLP seeks to address this, right at the source of the complaint, actually reading the complaint and extracting the areas of the organisations service that the customer is having difficulty with.

Secondly, NLP is looking to harness the emotion and sentiment that complaints tend to be rich in.  How does the customer feel. Are they angry?  Are they upset?  Are they likely to tell others of their dis-satisfaction?  Do they want a simple apology?  All of these are valid queries that can help drive automated decision making within a complaints process.  This is what NLP seeks to extract and provide.

Finally then, if RPA has scoured the wider public realm to identify dissatisfaction with service, NLP has automatically read the complaint and figured out what the problem is, then we can use Machine Learning to take the outputs from NLP and output predictive analysis on the steps that should be taken in order to address the customers complaint.  Automatically.  Just think about that for a second.  The learning model would know what the customer is complaining about, then learn over time for a particular organisation what actions should be taken to address the complaint.  The learning would get better and better as it trains on the complaints volume received.  Its appetite satisfied by our 24/7 robots trawling social media for any hint of dis-satisfaction.  

Welcome to complaints management 2019 - 2024.  It may be utopian to think that such a process could be fully automated, but in my eyes at least, it’s not unattainable.  With the technology now available to make it possible, the only difficulty may be running out of complaints...

Equiniti ICS work in partnership with public sector organisations across the UK and Ireland helping them to deliver of digital transformation objectives. If RPA sounds like an area you are keen to speak more about, contact us and we will be in touch. 

We have detected that you are in United States. We think that Equiniti US would be more suited to deal with your needs.