Many companies are declaring this year as “The Year of Automation” — seeing automation as an essential part of how we address our new normal.
Covered in this chapter
- Automation as an essential part of how we address our new normal
- Automation is designed to eliminating repetitive tasks and to augment humans’ skills
- The process of automation involves continuous discover, decide, act and optimize
- Twelve examples of business and IT automations
- Debunking the top three myths of automation technology
- Introducing the IBM Automation suite of products
The Art of Automation: Table of Contents
Here we go…
This introductory chapter to the Art of Automation provides the fuel that powers the chapters to come. Starting by establishing AI-powered Automation as a key aspect to the new normal emerging from the worldwide pandemic, you will learn about the process of automation and how it is structured around a feedback loop involving phases of discover, decide, act and optimize. Applying artificial intelligence (AI) during each of these phases propels automation to go beyond what was doable before, including understanding opportunities to predict versus react and see actionable patterns in noisy unstructured data.
This chapter outlines the styles of tasks that can be automated across an enterprise, while also clearing the air by debunking popular myths about AI and automation that readers need not be burdened with while reading the book.
“The Year of Automation”
Many companies are declaring this year as “The Year of Automation,” seeing automation as an essential part of how we address our new normal.  The COVID-19 pandemic has changed the dynamics of business and how we work.
For example, YouTube recently said in a blog post that with fewer people in its offices around the world, automation software is doing more of their video content moderation. “We have started relying more on technology to help with some of the work normally done by [content] reviewers,” the company said. “This means automated systems will start [curating] content without human review.”
The ideas around automation are simple, “tried and true” and have been applied to business since the very beginning of the industrial era. So, why do we feel the time is right now for automation?
Businesses today are all-in on being digital. The pandemic has accelerated digital transformation, with digital means to do just about everything from ordering a pizza to telemedicine to our Zoom-enabled workplaces — all digital-powered. As every aspect of a business becomes digital, the doors open for automation. In a sense, businesses, like cars, have become computers that can be programmed, and automation is the software that can propel a business to have “autopilot” or “auto-assist” modes. Digital-powered then sets the table for AI-powered.
Today, we have technology that can greatly change the “state of the art” of automation, with artificial intelligence (AI), machine learning, computer vision and natural language processing. AI has given birth to new ways to automate things that were difficult to automate before. AI, in the form of machine learning models, applied to IT automation with AIOps can predict the risk associated with making a change to your application, avoiding costly outages. Similarly, on the business-side, natural language understanding with robotic process automation can put automation in the hands of every business user to automate time-consuming and error-prone data entry tasks.
So, The Year of Automation is ushered in by a pandemic-accelerated, digital-powered revolution and is being paired with AI-powered technology, setting a big stage and placing intelligent automation at its very center. As such, many industry analysts have declared automation — or extreme automation, hyper-automation, etc. —as the most critical technology trend to act on now. 
Fundamental aspects of the Art of Automation
Automation as an “art form” for business could be simply viewed as a two-step dance. As discussed in the previous section, our new normal is driving client-demand to automate processes that eliminate repetitive and monotonous tasks. This is the first step in the dance and enables the second step, which is to augment humans to produce super-human results more rapidly.
Here is a closer look at these two simple yet fundamental aspects of the Art of Automation:
Task elimination targets simple, repetitive tasks across business and IT. Automating these tasks will free employees up to do more thoughtful work. For example, with closed offices keeping many of its workers away, PayPal has turned to chatbots, using them for a record 65% of message-based customer inquiries in recent weeks. “The resources we are able to deploy through AI are allowing us to be more flexible with our staff and prioritize their safety and well-being,” PayPal said in a statement.
Task augmentation supports, speeds up and increases employee efficiency. For example, with the increased use of online services during the coronavirus pandemic, AI-powered customer service agents can allow a single agent to help more users, decrease service queues and increase customer advocacy. AI is used to gauge user intent and capture information and the nature of the problem the customer is asking the company to solve. An automation workflow can then examine possible resolutions without engaging a human. That being said, the most powerful form of task augmentation is when humans and AI systems work hand-in-hand in achieving the desired outcome.
It is tempting to some businesses to stop after the first step and declare victory because of cost savings associated with task elimination. However, the true Art of Automation crescendos with augmenting humans with automation software. On Episode 3 of The Art of Automation podcast, Claus Jensen, Chief Digital Officer at Memorial Sloan Kettering Cancer Center, commented that automation software won’t replace doctors. Instead, it’s really the combination of artificial intelligence and human doctors which he stated, “outperforms either in isolation.” Claus further exemplified his comment with how doctors are using AI and computer imaging to enhance x-ray photos, highlighting anomalous structures, which then allows a doctor to discern and make a diagnosis, which would not have happened if the doctor used his eyes alone. A great example of the Art of Automation in action!
Autonomous vehicles and the process of automation
Automation has graduated to hyperautomation. We see it on the streets, with self-driving-autonomous vehicles. This is truly a marvel of modern ingenuity. Leveraging the best of artificial intelligence, internet of things and cloud computing, the automotive industry has set the “bar high” with this accomplishment.
Test driving a Tesla Model 3 with its autopilot features is a fun way to witness the process of automation at work, while gaining an appreciation for how Tesla has taken automation to an extreme with advanced technology.
Think about the “wow” moment when the car changes lanes by itself. This outcome is achieved through a process described as a set of tasks. This process of automation involves discovering data from sensors, including radar. As data is discovered, it is analyzed and correlated, such that another car might be recognized ahead. Then decisions are made. Is the car ahead too close? Based on a tolerance threshold, a proactive decision to trigger an action to slow down or switch lanes occurs. The discovery, decisions and actions occur simultaneously and are continually optimized against past and present data to predict future actions, forming an AI-fueled closed-loop automation system.
While there are dozens of variations of closed-loop automation (remember MAPE loops?), we prefer to use this simple 3 plus 1 methodology described in the above example; Discover, Decide, Act and Optimize.
Discover involves collecting, organizing and classifying structured and unstructured data that flow through your enterprise. Structured data is comprised of clearly defined data-types whose pattern makes them easily searchable. Unstructured data — “everything else” — is comprised of data that is usually not as easily searchable, including formats like audio, video and social media postings. 
AI is used to understand relationships and correlation, derive deep insights, find gaps and establish baseline KPIs. Patterns recognized in data can be classified as bottlenecks, hot spots, anomalies or outliers. They provide a context for decisions to optimize business and IT processes, using historical performance and predictive analytics to help you deal with variations. Automation occurs when decision-logic triggers actions that can proactively alert, tune configuration, make API calls and run programs that once required employee intervention. The “plus one” is optimize, which is happening throughout all these phases, providing a feedback loop that continually discovers new data patterns, improves decisions based on past actions and provides explanations and evidence that allow actions to be performed autonomously with confidence.
The autonomous vehicle example is inspirational. Now, how can we apply technology to similarly transform your enterprise with process automation and advanced technology to establish insights and automate actions?
AI-powered Automation is Enterprise Automation 2.0
To achieve these results, we are actively working to advance automation technology towards AI-powered Automation, which we declare to be Automation 2.0. AI-powered Automation is defined as a continuous closed-loop automation process where data patterns are discovered and analyzed, such that decisions on insights from the data can be translated into automated actions, with AI providing proactive optimizations during each stage of the process. AI-powered Automation uses actionable intelligence to deliver IT and business operations with speed, lower cost and improved user experience. The next section examines these four stages, illustrating how AI is transforming at each of these stages.
Better understand and classify unstructured data and processes so you can lessen the burden of manually analyzing and orchestrating actions.
Without AI, data discovery associated with automation is mostly limited to structured processes and structured data. Unstructured data is inherently noisy and usually slows down the automation process. With the use of machine learning (ML), models are produced to cut through, tease out and detect patterns in the noisy data. For example, with a properly trained classifier model, unstructured documents can be “structurally classified” as an invoice or insurance claim. Similarly, alerts from an IT system can be grouped and matched to a specific trouble ticket. With AI, the discovery process is no longer blocked by lack of structure; it uses AI intelligently to move from discovery to decision making.
For a deeper explanation of the nuances between different types of AI technologies, see “AI vs. Machine Learning vs. Deep Learning vs. Neural Networks: What’s the Difference?“
Combine the precision of IT automation with well-defined methodology of business automation so you can automate faster and with more accuracy in both IT and business.
AI-powered Automation aims to comprehensively provide a converged business and IT automation system that operates across a broad range of labor types, including business workers, solution architects, software engineers, IT operations, SRE, security and compliance engineers. By discovering data patterns across business and IT, decision-making can now be more impactful versus systems that are siloed to specific parts of an enterprise. An example of this is to correlate activity across software development and IT operations. In this case, changes to source code and configuration during development can be matched against incidents happening in a running IT system to predict risk associated with future changes to that code or configuration. By applying AI to automation, we are greatly improving the speed with which an enterprise can react to new patterns discovered.
Engage software bots more naturally and collaboratively so engagements become more self-service and productive.
The automation process is further differentiated in how automated actions are carried out. The gold standard in automating actions is robotic process automation (RPA) technology. With power from AI, we are evolving RPA from simple robotic scripts to becoming a tech that is more like a digital twin in the workplace. A digital twin is a virtual model of a process, product or service. This pairing of the virtual and physical worlds allows actions to be simulated in order to head off problems before they even occur, prevent downtime, and develop new opportunities. Furthermore, Automation 2.0 uses advanced natural language processing to produce a more collaborative relationship between AI and employees to produce a hybrid-workforce.
Predict potential incidents earlier so systems can proactively resolve issues before they impact normal operations.
Optimizations are continuously applied during discover, decision and action phases, capitalizing on new insights to autonomously enhance business and IT operations through closed-loop feedback. In Automation 2.0, optimizations move beyond reactive to predictive and proactive. With an end-to-end view of data across business and IT, AI-powered Automation can anticipate fluctuations and help avoid overreacting. For example, by combining structured and unstructured properties of historical change and incident records from enterprise IT, linkages can be extracted between change-incidents to create empirical evidence as new inputs to a change risk model. As new changes are being rolled out by IT, real-time proactive alerts can be issued based on predictions that illustrate why these changes are high-risk based on past evidence. Gartner Market Guide for AIOps platforms declares this proactive style of risk management as the most sophisticated stage of automation.
Putting the AI in AI-powered Automation
Machine learning and artificial intelligence (AI) enable new forms of intelligent automation. As the software “learns,” it becomes more adaptable. These technologies open the door for automation of higher-order tasks as well, rather than just basic, repetitive tasks. Automation is not just about automating those tasks humans are doing today; it’s also about realizing new potential opportunities.
As data sets become more thorough and available, and as software draws on more sources and synthesizes more data points, contextual information in human decision-making will only improve. Machine learning, then, will serve as a supplement to human knowledge. IBM Automation includes embedded Watson AI services to facilitate the liberal use of AI to make automation more effective across business and IT.
Examples of business and IT automations
There is strong evidence that there is demand for business and IT automation. Hundreds of thousands of discrete tasks make up the thousands of activities that drive the hundreds of processes within a digital enterprise; each individual task is an automation opportunity.
So, where does one begin? Developing an automation strategy in advance enables organizations to optimize investments by striking a balance between the difficulty of automating a task with its potential increase in efficiency. An IBM Institute for Business Value study showed that “one out of two executives using intelligent automation have identified the key processes within their organization that can be augmented or automated using AI capabilities.”
Analyzing work activities is the most accurate way to assess the potential for automation. The American Productivity and Quality Center (APQC) publishes a list of almost 1,100 cross-industry activities that compose 300 core enterprise processes. These processes hold the greatest potential to gain greatest efficiency.
One might correctly conclude from these studies that just about “anything and everything” is in play to be automated. That said, the following section takes a closer look at the sorts of tasks that can be automated in an enterprise across three specific classes of automations, which include ITOps, software delivery and business categories:
IT automation examples
A closer examination of 884 IT Ops automation use-cases from IBM Consulting illustrated four primary categories in labor can be offset by automation:
Service request management: Access provisioning, org changes, compliance requirements
Event management: Monitoring, alerting, remediation of commonly occurring events, self-service
Access management: Provisioning, revoking access, bulk access, onboarding/offboarding
Service desk and ticketing: Similar ticket identification, report generation
Software delivery automation examples
On the software delivery side, we see automation opportunities where gaps fall into these additional categories:
Application performance management: Hot spot analysis, impact analysis on environment (processor, CPU, memory)
Compliance management: Conformance to industry regulations (including NIST PCI, GDPR, HIPPA)
Application code vulnerability management: Open source code provenance
Software quality management: Anti-pattern detection, test coverage, release risk assessment
Business automation examples
From our business automation experience, we know automation gaps fall into four categories:
Workforce management: Email marketing, talent acquisition and employee recruitment, customer service chatbots
Case management: Route case ownership with queues, assign cases automatically, respond to customers automatically, escalate cases when necessary
Policy and compliance: Reporting compliance status and audit information, continuous verification of compliance requirements, manage risks and cache potential weaknesses
Process exception handling: Invoice processing, reconciliation, and approval
In the chapters to come in this book, these 12 categories of automation will be further explored both from a technology- and industry-usage perspective.
Debunking automation myths
Speed of automation
Myth: The automation process is a slow one.
Truth: The automation process of discover, decide, act and optimize might lead one to think that automation is a sequential and time-consuming process. While it is absolutely true that perfecting the automation process can take weeks or months, there are fast paths forward. For example, the use of RPA and low-code development are all designed to speed up the automation of “bite-size” activities or processes so customers can get immediate ROI without having to wait until the entire end-to-end process is automated. The fast turnaround time also allows business and IT to fail fast by iterating quickly and responding in real-time to external forces.
Automation for everybody
Myth: Automation is only for the likes of data scientists.
Truth: AI-powered Automation does not require everyone involved to be a data scientist. On the contrary, the use of AI enables automation technology to reach the general business user population, in addition to the IT developer, highly skilled knowledge workers and, of course, the data scientist. Users across the enterprise benefit from pre-trained models that were prepared in advance by experts — allowing for immediate use — without requiring deep AI skills. Delivering AI-powered Automations using natural language and chatbots creates an environment where the automation system meets users where they work, the way they work. This provides a more natural interaction, enabling more workers across an enterprise to both contribute and benefit from automation.
Automation and jobs
Myth: Automation will replace humans and take jobs away.
Truth: One of the greatest myths we hear about automation involves concerns that somehow software will replace humans and take away jobs. On the contrary, automation will add jobs. There are two key aspects to AI-powered Automation. Automation is about freeing up labor, such that workers can focus, on what matters most to business.
It’s about doing more meaningful work; work that will result in a better customer experience. It’s a shift of focus towards more productive work, not the removal of work. In fact, much of the evolution of technology goes in this path. The technology industry has always, created many more jobs, then it has eliminated.
IBM and the future of AI-powered Automation
The chapters to follow in this book are written by IBM subject matter experts in AI-powered Automation. As you will see, they don’t just “talk the talk,” they also “walk the walk” — meaning that much of their skill and experience in automation has been gained by working on automation products, technology and customer engagements.
As you read this book, you will sometimes hear references to products and technology that have influenced their point of view on AI-powered Automation. Therefore, this section provides a 30,000-foot view of the IBM Automation offerings to assist you in seeing how all their individual slices fit into the bigger pie. This section is not meant to be an IBM advertisement, but instead is here to provide a backdrop into how and why the technology chapters of this book are organized the way they are.
IBM’s approach to AI-powered Automation takes the form of a converged business and IT automation system with an ability to continually optimize by discovering, deciding and taking action as a means to automate processes across an enterprise. With this end-to-end view of automation, we are taking a bold step towards creating a hybrid workforce where your employees — collaborating with their digital twins — can gain deeper efficiencies across their business and free up time and money to focus on new business opportunities.
The IBM Automation offerings has three suites of capabilities:
- Business Automation
- IT Automation
The three offerings suites under IBM Automation are delivered as IBM Cloud Paks®, which are containerized software that uses Red Hat OpenShift as the means by which the software can be run and managed on any cloud — public or private.
Each Cloud Pak has best of breed automation unto its own. However, when the Cloud Paks are deployed together, their value is multiplied. For example, when Business Automation and AIOps are deployed together, business and IT events can be linked such that a business user can, in real-time, assess the business impact (e.g., time, cost, customers effected) by an IT incident. They can then use automation (RPA-bots) to notify impacted users of the situation or take corrective action to reduce impact before users even realize an anomaly occurred.
Business Automation: Providing technology solutions in the domain of business automation, targeting line of business users and the predominate user persona. These capabilities are included in the IBM Cloud Pak® for Business Automation, including the following:
- Process automation and orchestration (workflow)
- Decision processing with business rules
- Intelligent document processing
IT Automation: Providing technology solutions in the domain of IT Operations Management, targeting ITOps, DevOps and SREs as the primary user personas. These capabilities are included in the IBM Cloud Pak® for Watson AIOps, including the following:
- AI Operations (AIOps)
- Multicloud management (MCM)
- Observability and application performance management (APM)
Integration: Providing technology solutions in the domain of system integration, targeting integration engineers, ITOps and developers as the primary user personas.
These capabilities are included in the IBM Cloud Pak® for Integration, including the following:
- Application integration (App Connect)
- API technology (API Connect)
- Messaging and events (MQ and Event Streams)
Automation foundation: Each of these Cloud Paks use a common set of automation services called the IBM Automation foundation (IAF). IAF is the glue that binds together the IBM Automation offerings and provides the additive value as the Cloud Paks are used together. The following shared automation services are available in IAF:
- Embedded Watson for machine learning and natural language processing
- Event and data analytics cloud (Apache Kafka events, Flink streaming analytics)
- Process mining
- Robotic process automation (RPA)
IBM Automation is one way to explore many of the technologies that will be discussed in the chapters to come. So, if you are looking to getting “hands on experience” with robotic process automation, intelligent document processing, integration and APIs and AIOps, simply go to IBM Automation and give it a try.
This chapter has outlined the basic definition of AI-powered Automation, but we’re really just getting started. In the chapters to follow, we further define and examine the process behind AI-powered Automation by expanding the view to include a deeper dive into its key capabilities and architecture.
The chapters to come cover similar topics explored in episodes of our “The Art of Automation” podcast, where guests on the show are subject matter experts, sharing different aspects of AI-powered Automation. These very same subject matter experts have been invited to co-author chapters in this book and share their examples of how automation is changing everyday life for the better.
The following illustration provides a high-level map of the chapters which is an easy way to understand the structure of this book:
The book starts with a series of chapters, written by subject matter experts, that deep-dive in the areas of Business (chapters 2-5) and IT (chapters 6-8) automation.
Following are chapters that cover industry use-cases (9-14). These chapters take the form of enhanced transcripts from the actual episodes of the Art of Automation Podcast, and they feature the perspectives of podcast guests who share their experience with AI-powered Automation within the context of their industry.
Chapter 16 is a summary chapter that looks at automation past, present and future.
And last but certainly not least, we conclude with a fun look at the unique cover-art, of each podcast episode and the stories from the artists that created them.
The Art of Automation: Landing Page