What and How of Automation & Intelligent processing

Automation and intelligent processing are everywhere. Almost all machines are now filled with algorithms. The mobile that we carry is nearly a supercomputer that supports us in all walks of life – be it finance, health, weather, news, and sports. It also monitors our movement, preferences, and actions and feeds into a self-learning machine. The browser that we use knows what we have purchased and which advertisements to be shown to us.  

In offices and factories - a lot of processes are being automated. In banks and financial services credit analysis, monitoring, and loan processing have almost become automated. The manufacturing industry was anyways well ahead in this automation thinking. The schools have also experimented with robot tutors. In short automation and intelligent processing are really all-pervasive.  

However, not all automation attempts are successful. Even when they are successful they may not be always cost-effective. 

This article is my attempt to simplify the decision-making around automation efforts in our professional life – especially in the typical knowledge processing scenarios.

I will try to drive the learning through five examples. These examples are easy to understand. These will help us to arrive at the pointers for automation.

Case # 1 – designers not filling the work done in the log sheet

Imagine you are managing a creative unit of a design firm. You have hired a set of designers who work on various assignments of different clients on a typical day. Their efforts have to be charged to the client’s work order as their time has billing associated with it. However, you face resistance from them to fill out a log sheet/timesheet. You come up with various options of automation and manual processes to capture the correct time against the work orders.

You may come up with 10s of solutions to solve this problem. Automation is the most preferred option; it may involve screengrab, simple keystroke capture, and many other ideas.

Hold your thoughts. We will come back to this little later. For now, stay with the problem.

Case # 2 – pencil vs. the perfect pen for the outer space

This is a very popular example. I am not completely sure whether this is a real or an imaginary scenario. You may recall that – somebody had manufactured a pen that can be used in space (where gravity is non-existent). The counterargument of this automation/complex problem solving was the usage of simple-humble “Pencil”.

Pencil definitely works but then scientists had spent a lot of effort to come up with a “Pen” solution.

Hold your thoughts. We will come back to this little later. For now, stay with the problem.

Case # 3 – Weight calculation system on the conveyer belt – a fan was enough

This is again a very popular example. You may recall that in this a manufacturing unit had the problem with some food packets remaining unfilled when they finally land on the packing conveyer belt. The solution to finding out which packet remained unfilled with goodies was a complex one – involving automated weight calculation and heat mapping, etc.

The counter solution was a simple fan attached to the conveyer belt which can throw out (with the appropriate wind power) the empty packets off the belt.

You may appreciate the solution – but hold your thoughts. We will come back to this little later. For now, stay with the problem statement and both solutions.

Case # 4 – collaboration minutes / follow up

A typical office setup. A Big bank may be. Thousands of employees and vendor staff. 100s meetings. 500+ projects and products. A buzzing office life. You may end up attending 5 to 7 meetings in a day. Each meeting requires notes taking, actions, and risks to be discussed and allocated. What is the current solution? All minutes are typed and each action is manually tracked with PMO staff and other management efforts being spent. Reminders/ actions and all aspects are manually followed up before the next meeting.

Now you have started thinking about the automation of some of these tasks. Auto reminders, speech recognition, robot secretary, etc.

Hold your thoughts. We will come back to this little later. For now, stay with the problem.

Case # 5 – Sales (field staff) – data entry and all the way to customer fulfillment

Sales lifecycle and CRM are two of the many business processes which have got many companies interested in automation, machine learning, and AI. I see big names focusing their efforts on this and maybe solutions are just around the corner. However on the ground what I see are still huge gaps. When I visit a car showroom – the follow-up calls that I get (surprising maybe only 1) or continuous connect that is expected in this age of automation and algorithms are very poor.

The follow-up is literally non-existent.  And this is not just with car sales representatives. Even in the software industry. The post enquiring or proposal follow-up is literally non-existent. The SMS and emails are hardly effective with SPAM scaring away each and every person. 

Now you have started thinking about the automation of some of these tasks. Auto reminders, video demonstration, usage of WhatsApp, etc.

However, at the same time – the best sales experience that I have got is from a smiling salesperson, respect that I would have got, and really good advice. The human-to-human connections still work and are most effective. Don’t we hate the IVR system – and are desperate to speak to a reach “human” voice?

Before we summarize the pointers of automation let’s appreciate the advancement that the human race has achieved. The technological growth of intelligence that the human race has achieved is unparalleled. Each step towards new technology even though not immediately useful will be used in other scenarios (e.g. the pen that works in space would have found a new usage of that technology).

The computing power that we now have in our mobile phones was not imaginable just 20 years ago. 

Maybe we would reach singularity (where AI will actually be AI) eventually.  At the same time, problems still remain - the ecological imbalance, lopsided distribution of income, etc. 

Thus the best and most suited automation/ usage of computation is one that is well balanced. 

Now let’s summarise learning’s from 5 cases that we had seen earlier.

1.      Automation is not always the simplest or cost-effective solution (e.g. Pen or Fan)

2.      Automation can solve problems but are they preferred and cost-effective? (e.g. filling the log sheet is much easier and cheaper than any automation solution)

3.      Automation where needed has not reached that maturity (e.g. meetings or sales follow-up)

4.      Automation can’t yet replace human-to-human connection. (e.g. smiling and the courteous salesperson or human voice instead of IVR)

5.      Automation of simple processes such as data capture or speech recognition has still some way to go (e.g. minutes capture)

6.      At the same time automation and intelligence is already all-pervasive (e.g. mobile phones)

7.      Most manual data entry or simple logics/analysis will soon become fully automated (e.g. algo trades or news capture and then decision making/order suggestions in stock markets based on news)

8.      Many automation attempts are trying to solve the symptoms than the root cause (e.g. automation of filling of log sheets)

Summarising further; we can come up with the following takeaways:

Automation is possible and preferred for the following situations:

o  Simple processes can be automated  

o  Analysis and decision making (somewhat) can be automated

o  Data entry / capturing will get automated

o  Most human touchpoints with systems will automatically get automated e.g. our interactions with our mobile or browser

Automation is not possible as yet or not preferred:

o  Where system can’t make effective decisions or customer service e.g. car sales

o  Connecting to the grand scheme of things – a level of interconnectedness with more data points e.g. arranging an event (event management) can't be automated

Automation is not advisable (at least from the point of view of the team or organization attempting to automate)

o  Culture or Behavior should not be automated (e.g. people not filling timesheets or following data capture processes)’

o  Symptom should not be automated (e.g. accounting process can’t be automated to come up with accounting entry – proper accounting is still a domain of humans)

Hope you agree with the takeaways. Automation and intelligent processing are unstoppable phenomena - all we can do is channelize them better in our life, office, and our organization.