Telstra is looking to automate the way it finds and documents internal processes, which could allow it to locate opportunities for improvement much faster.
Head of intelligent automation Tim Barnes told the Blue Prism World conference overnight that automated discovery would supplement his team’s existing approach in manually scoping out processes that are deemed good candidates for robotic process automation (RPA).
So far, the intelligent automation centre of excellence (CoE), which Barnes leads, has automated 170 processes and delivered between 40,000 and 50,000 hours of staff time a year back to the telco’s different business units.
“We've certainly reached scale,” Barnes said.
“We've delivered bots into every functional unit within Telstra.
“Most of our work is now in our Global Business Services so in shared services: legal, procurement, activations, provisioning, and field services. That's where probably the majority of the opportunity is [to automate] boring, repetitive, manual work.
“We probably do 70 percent of our time in Global Business Services now.”
Barnes’ team engages with different parts of the Telstra business, offering a full suite of intelligent automation services and training, with the business unit picking up more of the workload over time.
“The whole premise of starting the CoE was to really empower the business - to take these capabilities around automation out to the business and teach them how to use these tools so that they can solve their own business problems,” Barnes said.
“Essentially we go in, we look across the business and see what the opportunity is within their pipeline.
“We then scope those opportunities out, write the relevant documentation, do the development for them, and support that on a centralised platform, using a centralised governance and design authority.”
While this had worked well so far, Barnes said the scoping work was manually intensive.
“When we first started doing automation from a CoE perspective, we would take an organisational diagnostic approach - essentially going in, meeting with the executives, and then doing almost a ‘speed dating’ of all of their processes, looking at all of their processes at a very high level, and really just assessing what the automation potential and the value of the work effort involved,” he said.
“Essentially we'd come up with a list of 40 or 50 potential opportunities. We’d look at what's going to create the most value, and how hard is it going to be to deliver, and then just pick them off.
“It's quite a simple approach.
Even pre-COVID, it was challenging for subject matter experts (SMEs) to come to terms with all possible variations of a process to then be in a position to work out which parts might be candidates for automation.
“Think about PSTN lines, for example,” he said.
“You can have thousands of different variations to that type of activation. No one's ever going to know all those different processes.”
Post-COVID, SME time is much harder to secure internally as they juggle competing priorities.
This has led Barnes’ team - and Telstra generally - to start experimenting with automated discovery tools for this work.
“As a big enterprise, we've been around for a long time and we have a lot of processes that have changed and morphed over the years,” Barnes said.
“Understanding our own processes and having them all documented is a challenge for every big enterprise, put it that way.
“We're really looking at tools now where we can automate the discovery of our processes, where we can document them at a work instruction level so we can really see where the variation is in each process and understand where the wastage is in a process.”
Barnes said he did not see automated discovery as “a silver bullet”.
“It's only going to get us partially there, but I think it will save ourselves and the business a lot of time,” he said.
“COVID has really impacted the business, and SME time has become a real premium as we work to keep services going for customers.
“Finding an automated way of discovering processes is really important so we can document them and we can really make sure that they're re-engineered if required.
“Fixing the processes up before you automate them is a really good idea.”
Barnes said that in the immediate future, process automations were likely to involve newer techniques that took advantage of machine learning, optical character recognition (OCR) and natural language processing (NLP).
“The ones that were really playing with at the moment are chatbots - I hate the name chatbot - but being able to use an automated way of responding to customers via text, voice or email is obviously a big area for us,” Barnes said.
“[It’s about] being able to respond to a customer at any given moment and then being able to pass that to a bot that can then fulfill the transaction once we're confident we understand what the customer requires.”
Meanwhile, OCR could be used “to understand and digitise some content ... from a PDF or an email and be able to apply some logic around that using NLP, which can then give you an idea of what the customer's intent is of what they're requesting.”
Similarly, that combination of techniques could be useful in taking a purchase order, “and then based on your machine learning capabilities, being able to understand who that's from, the invoice amount and then automate the management of that purchase order or invoice.”