What can you automate?

“If I had asked people what they wanted, they would have said faster horses”. The idea of a “car” much less a self-driving vehicle is unthinkable. We have come a long way with advancement in technology but there is no wisdom in relishing in our past technological successes. It is merely hindsight bias. On the flipside, anticipating future developments of technological development require foresight.

No one can predict the future developments. Afterall, the leaps in technologies we’ve experienced took time to evolve. However, what we can discuss is what can we do as businesses, to tap on existing technology to harness value.  

KR_MachineLearning-01-1

How can you identify tasks that can be easily automated with existing technology 

The Harvard Business Review suggests that each job should be deconstructed into smaller tasks before we identify specific tasks for automation. HBR segmented the characteristics of work into three spectrums: repetitive vs variable work, independent vs interactive work, physical vs mental work.  

 

1. Physical vs Mental (and digital) 

Automation of laborious work has been omni-present since the industrial revolution. Mental work which require human intelligence hasn’t been replaceable until recent years when machine are learning fast and artificial intelligence are developed. However, algorithms have chances of being biased, and should not be used to make important decisions like making judgement in the courtroom. Cognitive collaboration between man and machine is the ideal solution for the “best of both world”.  Tasks which primarily lives on the digital realm (such as sending emails) are more automatable than work in the real world (like packing the pantry).  

 

2. Independent vs Interactive  

Work that can be carried out independently are more automatable than work which requires collaboration between various parties. This is on the basis that interaction with human requires more communication skills and empathy.  

 

3. Repetitive vs Variable  

Repetitive jobs are mundane. Getting bored on the job can be indicative of the presence of highly repetitive procedures in their workflow, somewhat narrow scope of work, and the lack of challenge. Working on “boring” jobs hurts employee morale and businesses know that.  

This is great news for machines which are entities with no hedonic needs (meaning they don’t get bored). They are highly efficient with repetitive work, and excellent at performing jobs with simple decision rules.  

Variable work with limited variation is excellent petri dish for AI to grow its knowledge. However, highly variable work is less automatable. 

Simple ways to automate with existing technologies

Here, we share 2 ways that enterprises can automate work using the characteristics identified above

1. Cognitive Robotic Processing Automation (CRPA) 

RPA are suitable for task which are digital, independent and repetitive.  

RPA are lauded for automating repetitive, rule-based processes. This type of automation can be applied for administrative tasks - data entry, scheduled/trigger roles. Tools are available to schedule marketing posts and emails, to collect data in specified manners. Downside of RPA is that it requires high level of maintenance as the actions may be affected when there are changes to the interface.  

 

2. AI Chatbots - conversational solutions 

AI Chatbot are suitable for tasks which are digital, independent or interactive, repetitive tasks.  

Like RPA, chatbots are useful for automating onerous current processes – time consuming and difficult to navigate. It is also an ideal solution for tasks involving information which are easy to consume and control via conversation. On top of that, Chatbot excels for roles that requires human interaction, such as the customer-facing functions.  

AI allows Chatbot to be adaptable to task with limited variability as they learn. Companies with customer care team will benefit from implementing conversational solutions as chatbots will be able to hand-off to the live-agents when it picks up dissatisfaction from its end-users using sentiment analysis.  

 

How to tell if automation is viable 

1. Complexity of rules and standardization (SOP present)

2. Technology compatibility (api, integration)

3. Data availability (transcript, data available)

4. Firm readiness (security, accuracy, price, scalability) 

 

KR_Whatcanbeautomated-01

In the race, partner with machines. They are friends not foe.

For the employees 

There are bound to be worries about loss of jobs and likelihood and downsizing in the age of automation. Machines are friends, not enemy. You are not replaceable by machines. Machines help to increase productivity, so you can focus on the right things. Like with new friends, you may not know each other as well at the start and require some getting-to-know. It is uncomfortable to be left in a room with a stranger, but we will prevail. 

 

Closing thoughts 

At the end of the day, this could be the end goal for everyone – do only the most challenging parts of your tasks and checking on the quality of work produced, while work does itself. 

 

My dear fellow, if I have to do work, then I'm not doing my job. 

– Edward Darby, in Suits Season 3 

 

Related Resources: 

https://www2.deloitte.com/content/dam/Deloitte/cn/Documents/innovation/deloitte-cn-innovation-ai-whitepaper-en-190118.pdf 

https://hbr.org/2019/02/how-to-break-down-work-into-tasks-that-can-be-automated 

https://www.kofax.com/Blog/2018/april/robotic-process-automation-4-indispensable-types-of-robots-and-how-to-use-them 

https://www.cio.com/article/3236451/what-is-rpa-robotic-process-automation-explained.html 

https://www.mckinsey.com/business-functions/digital-mckinsey/our-insights/where-machines-could-replace-humans-and-where-they-cant-yet 

https://www.forbes.com/sites/jasonbloomberg/2018/11/06/why-you-should-think-twice-about-robotic-process-automation/#636338875fe1 

https://www2.deloitte.com/insights/us/en/deloitte-review/issue-20/augmented-intelligence-human-computer-collaboration.html