Point of View

Avoid the seven deadly IT automation sins to drive real outcomes

February 17, 2021

You might be relieved to read that we won’t nudge you to go confession to repent your automation sins. If anything, we want to foster a more realistic and relevant discussion on all things automation. With often bewildering noise levels around automation, it is not easy to find your bearings in a foggy technology landscape. On the one hand, we notice blurred perceptions of RPA (robotic process automation) are distorting discussions, even in IT scenarios where the technology has, at best, a moderate relevance. But broadly speaking, in the context of IT use cases, the term “RPA” is largely just a placeholder for unfocused automation discussions. On the other hand, “AIOps” (artificial intelligence in IT operations) describes a similarly broad category of innovation in IT operations. Yet, most use cases are confined to advanced analytics and suggestions of next-best actions rather than automation.

More importantly, these “automation” capabilities are generally confined to silos and point solutions. The overarching goals have to be much more holistic: Achieve the ability to deliver end-to-end, and decouple labor arbitrage from routine service delivery. Similarly, how do we achieve the cross-functional data strategies and workflows inherent in the OneOffice mindset? The intent is to overcome the organizational silos in front, middle, and back offices and, equally, between IT and business to truly enable digital experiences. As an industry, we need to drive the discussions on automation back to outcomes. To stay clear of purgatory, let’s take stock of the biggest automation sins, reflect on how market dynamics are changing, and learn what we can from it.

Exhibit 1: Focus on cross-functional data strategies and workflows to drive the IT services discussion back to outcomes

Source: HFS Research, 2021

Deadly Sin 1: Stop the IT versus business rhetoric—develop a OneOffice cross-functional mindset

The habitual finger-pointing when automation projects fail often tells a story about painfully slow progress toward aligning IT and business. Yet, citing a lack of IT support or pointing to ineffective stakeholder management ignores the elephant in the room. Only by making progress in overcoming organizational silos will organizations effectively support digital customer and employee experience outcomes. Adopting methodologies like DevOps or Distribute Agile enables change on the journey toward the OneOffice, but enterprises must complement these methodologies with automation approaches borne out of a cross-functional mindset.

So—what does “cross-functional automation” really mean? For starters, you have to understand and manage dependencies across functions. If an RPA project starts on the business side, yet management fails to understand the dependencies with infrastructure, applications, and quality assurance, there will be trouble in paradise. Once organizations progress with their journey, they have to embrace cross-functional workflows. I had expected for quite some time that we would see a convergence of IT and business requirements with tools—and it is finally happening. The oscillations of convergence are as disparate and complex, as service delivery backbones tend to be. We see moves to finally establish enterprise-wide service monitoring and service management; ServiceNow is increasingly the vanguard of those initiatives. Take the example of CPG firm Mondelez, which started its ServiceNow journey with IT services management (ITSM) but then expanded the platform across procurement and corporate legal. This is the journey toward the OneOffice crystalized.

Deadly Sin 2: Stop creating data stovepipes—make automation native to turn data into knowledge

A core competency for progressing toward the OneOffice is building out service delivery capabilities that enable organizations to automatically access, analyze, and store data to leverage it in dynamic workflows across different domains and organizational boundaries. These capabilities must surpass a narrow RPA mindset aiming to ingest predominantly structured data to automate it statically. Rather, processes must consume unstructured data such as IoT and Edge information or alerts from managing infrastructure and applications. Artificial intelligence (AI) is a critical enabler for deriving meaning from data and orchestrating and automating the disparate workflows that consume it.

If you take this cross-functional mindset as the North Star for operationalizing the OneOffice, it becomes abundantly clear that a piecemeal approach with a plethora of point solutions won’t get you closer to the promised land. My good chum Phil made this point succinctly by reinforcing our belief that automation is not your strategy: “Automation has to be native to any organization, giving it the ability to develop AI to orchestrate these processes. It is not ‘hyper’ or even ‘intelligent’…it just needs to be native.” However, these native automation capabilities are largely confined to silos and point solutions. The overarching goals must be much more holistic: Deliver end-to-end and decouple labor arbitrage from routine service delivery.

To achieve those outcomes, which are often tied to customer and employee experiences, we need to find ways of storing those vast amounts of data. Only then can we progress to reusing those data and information assets. The ultimate goal has to be knowledge transfer, which is pretty much the Holy Grail of AI. That is the ability to reuse knowledge, not just data, in different domains. To progress in that direction, the AI capabilities have to go beyond just machine learning, which represents largely statistical models. Fundamentally, these advanced AI capabilities need to orchestrate and ultimately automate knowledge across the various domains of the OneOffice.

Deadly Sin 3: Stop taking a compartmentalized approach to automation—cross-fertilize across RPA and AIOps

To progress toward those cross-functional workflows and inch closer to sharing knowledge across domains, we need to snap out of the comfort zone of traditional silos. We need to learn from experiences with different requirements and use cases. For instance, I find it baffling that when discussing cognitive automation capabilities, business executives often frown and point to compliance issues, particularly around explainable AI and ethics. Yet, the same organizations happily deploy similar tools across network and application monitoring and beyond. It often feels like we are reliving the same discussions we had around cloud many years back.

We are starting to see first glimpses of a convergence of IT and business requirements. RPA vendors like UiPath are pushing into test automation and supporting SAP automation, and AutomationAnywhere is partnering with Ayehu for AIOps capabilities, particularly self-remediation. Building on that, we see a cross-functional expansion of so-called AIOps tools. The AIOps landscape is highly fragmented, and the tools tend to be highly effective for narrow, specific requirements. This fragmentation  also means that most of these tools are not cross-functional, which is an essential characteristic for organizations overcoming their organizational silos. A first logical step for cross-functional requirements could be enterprise-wide service management and monitoring. Organizations need a single plane of glass that provides all the operational metrics and data. Through vendor consolidation, partnering, and integration, we are likely to see a horizontal shift that will enable those cross-functional capabilities. Another example of how this fragmentation can be overcome is Amelia (formerly IPsoft), which is driving complex process flows from the NLP interaction to the integration with ITSM solutions. At the same time the company is offering vertical end-to-end solutions that includes curated industry-specific knowledge, which allows Level 0 and 1 activities to be automated.

Deadly Sin 4: Stop running after the marketing buzz—embrace IT automation use cases that don’t always make the headlines

Surprisingly to some, there is life beyond RPA. And while you are asking—RPA is not the new ERP. To operationalize the OneOffice, organizations must integrate and orchestrate a broad and disparate set of tools and technologies. Operational leaders often struggle to segment all those approaches and, most importantly, prioritize relevant investments. We have tried to outline in Exhibit 1 the generic service delivery activities such as scheduling, monitoring, and alerting. But those activities must support a vast array of technologies and approaches. These delivery tasks are typically enabled by a plethora of point solutions, including automation; however, these solutions are deterministic and require enormous integration efforts. As such, they tend to be barriers in the journey toward end-to-end automation and the ultimate goal of the OneOffice. Therefore, organizations should embrace the automation flavors (see Exhibit 2) that are not yet making the headlines as essential for fostering a cross-functional mindset.

Exhibit 2: These up-and-coming automation flavors are essential for a cross-functional mindset

Deadly Sin 5: Stop using IT automation as a band-aid—progress toward cloud native.

Organizations must stop using IT automation (and RPA) as a Band-Aid for badly designed processes. Adding to that, the accelerated move toward the cloud necessitates new approaches to automation. As we discussed in an interview with our friends at Container Solutions, the journey toward being cloud native is more about culture, people, and process than it is about microservices, containers, and Kubernetes. The Holy Grail is a data-driven, collaborative work style underpinned by DevOps. Unsurprisingly, this is a far cry from the static and often brittle processes that we have today and where RPA has become the North Star. Thus, organizations must plan for this new reality and prepare for the next shift, which will be characterized by distributed, self-organized processes supported by preventive and even anticipatory AI interventions.

Consequently, the holistic approach that we described in Figure 1 will become mandatory for reaping the benefits of the cloud-native culture. A central element in that approach will be progressing toward self-healing automation. Only then will organizations move to a truly digital delivery model that can respond to changes in the environment in close to real-time. We will dive deeper into that by providing a litmus test for where we are with self-healing. Much of what is proclaimed to be self-remediation or even self-healing is instead insight that suggests next-best actions.

Deadly Sin 6: Stop using automation for just deterministic use cases—make it adaptive and autonomous

Picking up on Phil’s point that AI needs to orchestrate the automated processes, it needs to accomplish that in a way that is neither static nor deterministic. Thus, traditional automation approaches, often built around runbooks, are not sufficient because the scripts in runbooks need to be maintained, often manually. More fundamentally, however, they can only deal with defined problems. Therefore, organizations incur significant costs to maintain those runbooks, but runbooks’ deterministic nature means that their automations won’t adapt to environment changes. Yet, equally you don’t just want alerts that guide you to look up problems and solutions in GitHub. Here, the need for cross-functional AIOps and automation approaches cuts in.

To bring all those disparate automation technologies together, organizations need to evaluate technologies like arago. At the heart of arago’s thinking is a non-deterministic approach to automation enabled by new classes of algorithms and a platform that blends disparate approaches of AI. By having general problem-solving AI and not just pre-optimized machine learning solutions, arago’s HIRO platform can integrate complex, disparate data sets and adapt to changes in the environment. Thus, organizations could automate the automation as depicted in Figure 1 and overcome the limitations of runbooks and the narrow scope of AIOps. Crucially, the platform can deal with both IT and business requirements, thus enabling cross-functional workflows.

Deadly Sin 7: Stop looking for the panacea—embrace the mindset of the OneOffice Emerging Tech Platform

Of all the sins, believing the marketing hype that individual shiny toys will lead you to process nirvana is literally the cardinal sin. The context for all those discussions should be service delivery and the complexity of managing the journey toward the OneOffice. As my colleague Saurabh tends to put it, it is the power of AND. It is about integrating and orchestrating disparate capabilities. However, the crucial question is whether organizations should build out those capabilities organically or leverage service providers’ automation platforms. Platforms such as Accenture’s myWizard or HCL’s DryICE are brilliant pieces of engineering that cover the whole software development lifecycle. Yet, organizations have to be clear about who owns the IP and data and how portable created assets are.

Describing all the seven deadly sins is meant to call out that you need a mindset shift to support collaborative cross-functional enterprise operations powered by an integrated stack of emerging tech that complements your core and natively automates your processes. That is what HFS has termed the OneOffice Emerging Tech Platform, which will be a strong focus of our research in 2021. The tooling that sits within the stack is a representative set of emerging tech designed to function in integrated layers on top of existing investments in enterprise OneOffice applications. Something we’ve said before can serve as a summary of our deliberations: “Automation is a necessary discipline to make OneOffice function. It is NATIVE to OneOffice. You have to design processes to achieve the outcomes and data you need in the cloud. They HAVE to be automated to make that possible (whether through RPA, APIs, customized code, or off-the-shelf software). Once automated, AI can be APPLIED to orchestrate them to create the data we need to make rapid business decisions.”

The Bottom Line: Automation needs to be a core part of your IT modernization strategy with a focus on driving outcomes. It is core to our IT services research agenda.

Organizations have to cut through the marketing hype and focus their service delivery strategies on outcomes, not technology capabilities. The pandemic’s painful lesson is that the goal must be an end-to-end delivery of services. While task automation and employee productivity are excellent tactical efficiency gains, the focus should be on a holistic approach that sets organizations on course for cross-functional automation.

Exhibit 3: Clusters of the HFS IT Services research agenda

Source: HFS Research 2020

Our research agenda in IT services will reflect these issues as Exhibit 3 outlines. We will guide organizations on their journey toward Cloud Native. As with Distributed Agile, the key issues are about culture, people, and processes. To make those issues work in complex projects such as outsourcing is hard work. You will also see us focusing on quality assurance aiming to guarantee those outcomes that we have described. Equally how is innovation enhancing the respective assurance offerings? Lastly, high on the agenda and very close to my heart, you will see us expanding our automation coverage toward more IT-centric requirements with non-deterministic approaches that move beyond runbooks.

Sign in to view or download this research.


Lost your password?


Insight. Inspiration. Impact.

Register now for immediate access of HFS' research, data and forward looking trends.

Get Started