Automate IT...

Automation has moved beyond hype or a buzz word in today's progressive organizations. Though you may or may not hear it in same breath along with Cloud, Artificial Intelligence (AI), Machine Learning (ML), Big Data and Blockchain. The reality is that these all are being embraced, here to stay and are impacting us today. It is also true that all of these are evolving and have a varied influence on organizational strategies. Some companies are ahead of others when it comes to embracing these new technologies and techniques. All of these have a common underlying goal - "Automate IT".

Coming to the topic "Automation" - this is a technique to reduce human touch point. Industrial automation happened in 1970's - where robots replaced humans. How can it be applied to the world of Computers & Data? With Machine Learning, AI, Cognitive Capabilities being built into machines more and more there is still a significant numbers of decisions that humans make everyday. Some are repetitive in nature while some require additional cognitive capabilities. To automate a large stack of end to end operations, the systems implemented will use multiple techniques including finding ways to automate those decisions that humans make.

Blockchain is a technology built on cryptography to process information as it moves from one state to another. Cryptocurrency is one implementation of Blockchain and it has the potential to be embraced in supply chain, financial transactions and even real estate transactions where you can persist an entity and capture the changing states of the entity as validated or verified blocks by peer network that can be trusted without human intervention. There are many other evolving use cases of this technology which has the potential of removing humans from validation acts - example a teller to validate what is in your account or a title company to validate who owns a particular property.

Though Big Data is a technique to process large data sets however it has become synonymous with Hadoop and Hadoop implementations. Machine Learning is what can be accomplished with having large data sets where systems can identify patterns efficiently. You use analytical algorithms to understand these patterns and insights about your data with higher confidence. ML & AI go hand in hand, these additional insights provides us information to define rules and thresholds to enable the systems the make decisions than a human making a decision. This capability where machines can make a decision is referred as AI.

The world of AI is evolving more and more with the machines gaining cognitive capabilities. Cognitive capabilities refers to the capability of a machine to self correct as they learn more, means as they learn more their decisions can change. With more learning the machines gain higher confidence in their analytical capabilities enabling the machines to make more automated decisions. This capability by means is not perfect but growing exponentially in its capability. It takes time for the organization to take advantage of these implementations as for the machines too there is a learning curve, you have to keep feeding more data and decisions to the machines before the decisions become good. IBM's Watson is a cognitive AI system. The early implementation of these are being used in Insurance and Credit industries. Stock market also uses AI techniques where buy and sell orders for large investment companies are automatically triggered by machines than human intervention. Future use of this technology are plenty and how we get medical care today can be significantly disrupted with AI in our future.

Cloud is very diverse, it can be platform play, software acquisition play or be your new data center strategy. Some companies are migrating to cloud from their physical data centers because of better capabilities the cloud implementations are providing. More and more S/W companies have embraced cloud first strategy and their cloud implementation usually see new features first. Whether you are a small company, a start up or a large organization, you cannot ignore cloud. For smaller companies and startup's its a no-brainer as it requires minimal capital investment to get started. For large organizations this is an efficiency play, better capabilities better up time and reduced costs. Let us not forget faster time to market too. Large Organizations however need to have a cloud migration strategy in place than just get started. What cloud also enables you is to automate some of the core data center capabilities like self provisioning of additional hardware to meet peak load.

The repetitive processes where human touch point is required can be very easily automated using a technique called "Robotic Process Automation" (RPA). The concept of RPA as the name indicates includes a Robotic Agent which is a deployable software on a machine. Each agent can handle the task that 4 to 5 humans do. The tasks RPA agents perform are small scripts to replace what humans do. These scripts are small meaningful actions that a human may do. This requires a good understanding as well as documentation of rules, tasks and decisions. The human workloads decrease as the scripts usually are enhanced and made more complex. With time the human workload is deemed negligible.

There are multiple RPA S/W providers out there, the early one's with more mature capabilities are: Blue Prism, UIPath and Automation Anywhere. There are three types of costs:

1. Initial Setup Cost - Cost to acquire software, install software and the required hardware.
2. Cost to develop agents - includes cost of building initial scripts, testing, refining and implementing initial scripts called RPA Bots.
3. Maintenance and Support Costs - includes cost to monitor, learn and enhance scripts.

RPA can be implemented irrespective of your Cloud strategy or Big Data Strategy. The RPA vendors are making their software very smart and are enabling them with capabilities to leverage AI for automation. This capability should be on your evaluation criteria.

For any automation, change management is key to success. There is a underlying fear that jobs that humans perform will go away but the key is to help folks understand that new types of jobs will emerge. This is not a head count reduction opportunity but to become efficient in our operations by reducing human errors like fat fingering etc., introducing time predictability into operations and only leveraging humans where meaningful decisions need to be made. Additional types of jobs like business rules analyst, automation script writers, governance & audit and jobs to maintain/ monitor automation agents will also evolve as a result of this. We will see the need of data engineers and data SME's with deep industry knowledge to be in greater demand.

On a lighter note, I still believe humans are better at wine tasting or being a food or movie critic :-). As I type this I also believe I probably have jinxed this and someone in their lab are testing some algorithms to prove me wrong.

I welcome your feedback and questions.

Update 8/13/2019: Having done RPA for last 2+ years and thousands of bots in production - look out for an article/post on hype of RPA which will include do's and don'ts. What would you also like to see in that article? Please advise...