Ceox InView Issue 2
Insight and Tips on Digital Transformation, Case Management and Microsoft Technologies for IT & Business Leaders in the Public Sector and Social Housing
Has Intelligent Automation Come of Age?
Welcome to our Autumn 2020 issue of Ceox InView, written as we approach a winter of Covid-19 imposed restrictions and ‘new normal’ ways of working. As a reminder, Ceox InView is our way of providing insight and tips to help organisations navigate the world of Digital Transformation, Case Management and Microsoft Technology.
In this issue we explain why we think ‘Intelligent Automation has come of age’. In our first article we ask whether Intelligent Automation technology has got to the point where it is ready for use within the Public Sector. Our second article then looks at our experience of using Artificial Intelligence to help an organisation deal with increasing citizen enquiries. The solution automated the processing of incoming enquiries by extracting specific data and then routing it to the correct department for answering. A real-world example of Intelligent Automation in action delivering tangible efficiencies and benefits.
We hope you find this issue interesting and informative, please contact us via the usual channels if you have any feedback.
In September this year, Gartner published a paper stating that ‘AI is starting to deliver on its potential and its benefits for businesses are becoming a reality’. In addition, reactions to the Covid-19 pandemic have accelerated digital adoption to the point where 55% of customer interactions are digital. This has seen an increasing appetite from Public Sector organisations to utilise Intelligent Automation to deliver critical public services.
Intelligent Automation is using the combined power of Artificial Intelligence (AI) and Robotic Process Automation (RPA) to deliver efficiencies and automate mundane tasks. Often confused and used interchangeably, AI is having computers make decisions or perceive their environment. AI solutions need to be 'taught' so they have enough data to make a decision. RPA is where the computer follows a series of pre-configured steps, or a workflow. RPA is valuable for removing mundane tasks from a person's job, enabling them to concentrate on higher value activities. The real value comes when you combine these two, into Intelligent Automation, where a workflow can be configured with the computer also making decisions based on what it has learnt.
AI and RPA technology is not new, it's been around for many years and yet hasn't been widely adopted in the Public and Social Housing Sectors. With both of these sectors we see that technology needs to meet the following requirements before it's adopted:
Future Proof - It needs to be future proof and sustainable; organisations don't back technologies which need replacing every 2-3 years.
Technology Aligned – The technology needs to align with the existing technology stack and investments the organisation has already made.
Return on Investment - It needs to have a proven Return on Investment (ROI); organisations need to deliver efficiencies and realise cost savings from their investment.
Secure - It must be secure and compliant, enabling organisations to keep individuals' data safe.
This isn't the same in other sectors where for instance, financial services companies are able to spend millions on emerging technologies to gain a competitive advantage. Most Public Sector organisations need to take a more considered approach to digital transformation using proven (and secure) technologies to deliver value for money.
Over the last few years, Microsoft, through their Democratising AI strategy, have provided cloud-based services which enable any organisation to build an Intelligent Automation solution without the significant investment which would otherwise be required. Indeed, a number of years ago Microsoft introduced Power Platform, a suite of technologies which included Power Automate to deliver RPA functionality and AI Builder to add AI capability to applications. AI Builder provides several prebuilt and customisable AI models including Form Processing, Object Detection in images and Prediction models.
Power Platform holds data in a single repository (the Dataverse), ensuring that regardless of how the data is accessed, only the appropriate authenticated people can get access to the information. Sitting alongside Office 365 and Dynamics 365, Power Platform provides an accessible and inexpensive way to use Artificial Intelligence in real world scenarios.
So, it seems, that Intelligent Automation has come of age and 2021 is likely to be the year that Public Sector organisations start to deliver solutions with real benefit. Indeed, Power Platform seems to meet our 4 requirements:
Future Proof - Microsoft have created a long-term platform which they're throwing their weight behind. Microsoft are positioning Power Platform as their key business applications platform and have long-term plans to grow its capability.
Technology Aligned – Organisations who have already invested in Office 365 will find it very easy to add-on Power Platform services and the user experience will be familiar for users.
Return on Investment - The ready to use technology doesn’t involve a large upfront investment and integrates into the existing technology stack. Indeed, for a low 5-figure sum, Public Sector organisations can implement solutions which can have a significant impact on improving productivity and customer satisfaction.
Secure – The platform has been designed with security and manageability built-in so that IT departments can control solutions to ensure they are secure and safe.
All of which provides an opportunity for public sector organisations to secure a cost-efficient way to deliver Intelligent Automation.
“Between December 2019 and July 2020 digital customer interactions in Europe increased from 32% to 55%. A digital adoption acceleration of 3 years.”
McKinsey and Company - How COVID-19 has pushed companies over the technology tipping point—and transformed business forever. (Oct 5th 2020)
At Ceox we have just finished a project for a Public Sector client which used Microsoft's AI Builder, part of the Power Platform suite of tools, to categorise information. The project is an example of where powerful Artificial Intelligence (AI) is used to deliver real world benefits.
First, a bit of background to the project, the organisation receives large volumes of incoming data, sent by customers, in a free text format. They need to categorise this data, extracting specific information, before processing. The information can come in several formats, including website submissions, emails and letters. This scenario will be familiar to many Public Sector and Social Housing organisations, who can receive data for many different reasons, and need to then categorise that data.
On the project we used two components of Microsoft’s Power Platform, Power Automate to provide Robotic Process Automation (RPA) and AI Builder to provide Artificial Intelligence (AI). This enables an unattended (no one has to sit and watch) way of processing the submissions. We processed the submissions using two of the available AI Models within AI Builder:
Entity Extraction - This involves training the model to recognise specific data within a piece of text. In this instance we trained the model to look for the Recipient Name, Sender Name and Sender's Contact Details, amongst others.
Category Classification - This was used to understand the contents of the submission and put it into a corresponding category so that it can be processed by the relevant department. The model provides a suggested category as well as a confidence rating. If the confidence rating is low, then the automation redirects it to a person for them to decide on how to categorise it.
There were a number of things we learnt from developing the solution. Firstly, we established that when training the Category Classification model (teaching the AI how we’d like it to behave) required a lot less examples than we expected; for instance only 100 submissions where required to train each category of the model.
Working closely with Microsoft we identified that we only required enough training examples to gain a good confidence rating in the decisions. However, the plan is to train the model again once 3 months' worth of data has been collected.
Secondly, training only required the relevant piece of information. For instance, someone might indicate in their first paragraph what category their submission should ultimately end up in, so this is the text that was used in the training. This resulted in a higher confidence rating than if we used the whole submission for training.
The big drive for this project was to save the hours spent reading submissions and categorising them. It can be seen therefore that this processing saves hours of time for individuals who previously had to read the submissions, re-key the sender details and then assign to a relevant department. As the number of submissions increases the value of Intelligent Automation also increases as there is no extra cost for processing the additional submissions.
“Assuming an average processing time of 10 minutes per submission, if an organisation deals with 10,000 submissions a year, implementing Intelligent Automation can save over 220 days of effort per year.”
Whilst this solution was built for one organisation categorising information for one purpose, the underlying technology can be easily built upon and extended for other scenarios. For instance, service enquiries, complaints handling, application sifting, query processing and case management.
Intelligent Automation has come of age and can really make organisations more productive, removing mundane tasks and freeing teams up to concentrate on more valuable activities.