From Desktop Automation to Artificial Intelligence - A journey

In my previous post i.e. Do you really need RPA in your Business? I stated the high level criteria for any enterprise automation seeker to classify  a process as eligible for automation or not. 

Automation can also be classified into various stages. Some process automation only require partial automation or automation of certain steps within a workflow which is now popular by the name of desktop automation or RDA (Robotic Desktop Automation), Where as other business cases are more of enabling the organization wide systems with power to take autonomous rule based decisions which is popularly know as AI or Artificial Intelligence. 

The objective of this post is to understand the stages in the journey from RDA to AI and what are the criteria to achieve each state. 

1.     RDA (Robotic Desktop Automation):

The basic need for any enterprise that is very new to automation is that it wants to reduce the redundant work for its work force or increase the output of the workforce. This need is generally answered b automation of certain steps in the tasks which are conducted by the workflow. These steps may still need human intervention as the decision making is done by human beings. 

• RDA is process driven
• Manual intervention is needed as trigger is human initiated
• Decision making is done my humans
• Some steps in a workflow are automated to reduce tasks which were earlier done by human beings
• Examples are copy pasting data from excel to form or vice versa

2.     RPA (Robotic Process Automation): 

The next obvious step is to allow the process to take decision to trigger the event so that human intervention is reduced even further. Thus this level of automation intends to automate not just few steps but move towards increasing the level of automation in the entire workflow or process to achieve a final state of complete process automated with minimal or no human intervention

• ·          RPA is also process driven
• ·          Triggers are digital and self serviced
• ·          Intent is to automate the end to end workflow
• ·          Triggers are rule based
• ·          Examples are digital customer journey for opening of a new bank account

3.     ML (Machine Learning):

The next successive state for any enterprise in this journey is to allow the system do make decsions based on previous decisions made by humans. The in the previous two states we have seen that the triggers or decision are either made by humans or based on rules made by humans. This state however takes it to a new level where the decisions are made based on the data and decision rules are selected by machines based on the data.

• ·          ML is data driven
• ·          Perspective analysis and decision engines are at the heart of this stage
• ·          Triggers are based on historic data, rules are made based on historic data
• ·          Its not focused on one workflow but is capable of enterprise wide transformation
• ·          Examples are running decision engines and data analysis on customer journey’s and repeat purchase to come up with decision on product positioning and marketing mix

4.     AI (Artificial Intelligence):

The ultimate state for an enterprise is to be able to create intelligent systems which can not only make decisions but are also able to assist its stakeholders in making intelligent decisions. Intelligent systems can not only make decisions based on historical data but also are capable to deducing future events based on current set of steps and information from past.

• ·          AI data driven
• ·          Deductive analysis is the heart of this stage
• ·          System is intelligent to guide humans to take certain steps
• ·          Examples which I can think of “Jarvis” from Age of Ultron

Do you really need RPA in your Business?

Irrespective of which business vertical you are working in, process automation, machine learning and artificial intelligence are the words which you would come across day in and day out. Every business today is looking for ways to reduce cost and maximise revenue by using these technologies. However not all business processes and operations are fit for automation. This articles intends to cover the aspects which should be considered before jumping to conclusions on implementing automation for a particular business process.

Do I really need it?

• Would this automation reduce man hours? The main aim of any RPA implementation is to reduce the human efforts and increase efficiency. Thus an business which is considering RPA as a solution should work out n th number of man hours this implementation would reduce. The more the merrier. This can also be in terms of number of employees which are currently deployed to complete the process.

• Is the process based on business rules or on human judgement? Processes which are based on business rules are ideal candidates for automation. For example “if customer has provided a copy of the passport then issue air tickets else not”. Rule based processes can be converted into algorithms high can be understood by bots. Some processes may or may not be fully business rules driven and thus may be good candidates for attended automation.

• Are human inputs in binary formats or standard templates? Machines understand the language of 0 or1. If the process requires standard human inputs or human inputs are captured in standard formats such as forms, then the process is a good candidate for automation. On the other hand of inputs are non standard and require human judgement, such processes are not good candidates for automation.

Other factors? Other factors which determine the feasibility of RPA implementation include factors like how soon the process under consideration is expected to change.  Processes which are expected to change in next couple  to  months are not good candidates for RPA. Also if the number of exceptions are high in a process then it would need human intervention and thus RPA implementation would not be as fruitful/

Conclusion : RPA technology has huge potential for all industry sectors but businesses need to take a cautious decision while choosing which processes should be automated 

From Mass Production to Mass Customization : Data to the Rescue

Challenge:

Today since we are at the cusp of digital transformation across every possible interaction with a product and service offering, it is becoming more clear that digital transformation has started disrupting the affection of standardization across industries. Consumer today wants “tailor made” interaction with any product or service and industry want to achieve the cost benefit of with a standardization. These two are opposite forces, their fore  “customization” today is expensive and “standardization or mass production” causes customer dissatisfaction. 

Winners in this era would be those players who would be able to integrate these two i.e. “Mass Production” + “Customization” = “Mass Customization”. 

Deck:

Digital transformation of businesses, wide spread adoption of sensors and analytics are today transforming every industry. There was an era starting mid 1920s when the term Mass Production was popularized on correspondence with Ford Motor Company (Source). The idea was to mass produce using assembly lines to get cost benefits. Same was applied to services where standardized services such as insurance products, banking products, telecom plans, bus fares, road tolls were offered on a standard rates.

The emerging digital transformation and ever increasing use of sensors today is indicating that future would be flooded with data and insights. These insights derived from data gather by various sources of data such as sensors, data gathering software would help organizations gain insights into consumer behavior and thus offer “Mass Customization”.