One of the biggest trends right now is “leveraging” or “scaling” RPA’s reach by adding a cognitive element to the process.
Nadin Kraus from Teva Pharma, expert speaker at the 4th AI & RPA World Summit 2019, reveals why you should have an eye on this trend. Read the full report.
While RPA’s growth trajectory has been fairly steep – implementations jumping from 9% at the start of 2018 to 32% at the start of 2019 – there are limitations to what it can do on its own and organizations are running into hurdles in trying to scale.
Those under pressure to do more with less are keen to tap into new levers to drive automation’s impact further.
One of those levers is cognitive automation.
“Cognitive” steps in where RPA stops. It enables RPA to continue or expand its reach via two critical steps:
- Turning the unstructured data that RPA cannot handle into structured data; and
- Determining next optimal steps post RPA processing – i.e., using RPA outputs to drive additional, automated activities that require a human-like (i.e., cognitive) ability to act.
According to a survey conducted by SSON last year, half of the enterprises planning to implement RPA are also investing in cognitive capabilities.
Awareness is growing, in other words, but we are still in the very early days.
What is Cognitive Automation?
“Cognitive” implies automated activity that can think and act “like a human” – in other words mimic a human’s ability to view data and draw appropriate conclusions as to next steps.
Cognitive technologies can include speech recognition, natural language processing, and machine learning, which drive perception and judgment-based activities.
Cognitive relates not to output [predictive or prescriptive] but also to the way in which data is handled to drive optimal outputs.
Cognitive automation is gaining a lot of interest for content-centric processes where unstructured content is common. Some examples of cognitive solutions include Neuro Linguistic Processing, Computer Vision, and Machine Learning.
(Get the free accompanying report Cognitive Solutions Leverage the Power of RPA here: http://bit.ly/2IJtO25)
Teva: Preparing for Cognitive Advances Today
While most organizations are still in the early years of RPA implementation, vendors are advancing rapidly in the area of cognitive solutions.
Innovative managers like Nadin Kraus, Senior Director, Finance Innovation Europe, at Teva Pharmaceuticals, don’t want to wait until the RPA options are exhausted before considering what’s next.
Instead, she is already preparing now to get ahead of the incoming technologies.
“We started our RPA journey in April 2018 and are now at the stage where we are focusing on scaling up,” she explains. “There’s a long path ahead, still, with lots of opportunity for RPA. There is far more demand than capacity to fill, at present.”
However, Ms Kraus didn’t want to wait and possibly miss out on emerging cognitive opportunities. As a result, her team is already working on a Machine Learning solution to be positioned in front of the current RPA activity, to provide data needed to feed the automation.
“One challenge we ran into is that RPA works only for routine tasks, but often we need to see the business context and take appropriate decisions,” she explains.
The current RPA supports inventory valuation and reporting on write-offs by automating what used to be a highly manual process: logging on to SAP; defining parameters; running programs; downloading the results; and reporting these via Excel.
“Our RPA solution now runs all of this – but also reports through a new dashboard tool for our business users,” Ms Kraus explains.
While the RPA element works well, the fact that RPA works only for routine tasks is limiting its capacity, she says. “Our objective is to position Machine Learning in combination with the bot to analyze data in order to define parameters for write-off scenarios and drive predictive reporting on future write-offs to share with the business,” says Ms. Kraus. “This will allow the business to take preventative action.”
The challenge is that much of this logic is currently held in employees’ heads, she explains. It’s neither all easily accessible nor structured. “Machine Learning will change this.”
So how will this work?
The Machine Learning algorithm is being fed by historic data in order to predict future values of a specific parameter. “In our case, the devaluation of inventory on SKU level,” explains Ms. Kraus.
“If you want to rate your inventory in a detailed way, you need to know how many devaluations per SKU you will have in the next months. We are talking about more than 10,000 SKUs (for one country only) so that a detailed rating of the devaluation for each SKU is simply not performable by a human.”
At the moment, she adds, people are predicting future devaluations on basis of their instincts. For each devaluation, however, there are hundreds of relevant input factors that all have a certain percentage impact on the end result. “This logic is not really understood by humans,” adds Ms Kraus, “as it´s represented in our gut feeling.”
ML will identify these patterns quickly, however. “Based on historic devaluation data, the ML-algorithm will learn the knowledge and dependencies that are stored in the data. This knowledge will be used by the ML solution to predict future values for each SKU.
Together with our internal customers we can then define counteractions to prevent write offs, which the system will then propose based on each SKU case.”
Teva’s Finance team does not have the extra budget right now to invest in new solutions, while RPA is still delivering – but it has dealt with this by partnering with a local university that specializes in automation.
“One of our team is doing a Masters’ Thesis in cognitive computing at Ulm University,” Ms Kraus says. “So, we are supporting that Thesis, but at the same time the university is supporting us with its data scientists to train our resources. We are using standard software we already had in-house to prepare the relevant data and program specific ML algorithms.”
There is simply too much work for humans to stay on top of all this data, she explains. Particularly at month end, it’s nigh impossible for Teva’s Finance team to stay on top of 10,000 SKUs to assess write-offs for even just one country.
The exciting opportunity, Ms Kraus explains, is that her team will be able to not only define the appropriate parameters for assessing inventory write-offs, and also predict future write-offs and share this insight with the business.
“Having this level of reliable predictive data will help us in reducing write-offs, which is a costly expense in the pharma industry. RPA is helping to speed up and automate the process – but Machine Learning will leverage this performance by delivering better data and predictive insights.”
Learn more about this or similar topics by reaching out now. Contact Steven.Zapata@iqpc.de and attend the 4th AI & Robotic Process Automation WorldSummit 2019. Get more information at https://robotics-automation.iqpc.de/
Get the free accompanying report Cognitive Solutions Leverage the Power of RPA here: http://bit.ly/2IJtO25
Meet Nadin Kraus at our 4th AI & RPA World Summit 2019. Nadin Kraus, Senior Director Finance Innovation Europe, Teva.
Pharmaceutical / Ratiopharm will speak at the conference about the following topic:
Stream | Advanced Automation
Interview | Next Level Accounting – Hybrid Models for Intelligent Automation
- Which back-end processes to automate
- Virtual Assistants
- Automatic reporting
About the author: Barbara Hodge is Principal Analyst and Global Digital Content Editor of the Shared Services & Outsourcing Network (SSON), the largest and most respected forum for executives tasked with promoting and delivering optimized business services.