With all the headlines that have been published about the role of artificial intelligence (AI) in the cable operator community, it would be easy to conclude that we are on a rapid path to AI for IT operations (AIOps) ubiquity.
Alas, the hype may be a bit ahead of the reality on the ground. Despite significant -- and growing -- interest in how AI can be harnessed to supercharge network operations and accelerate go-to-market initiatives, I would rank the sector at a 3 or 4 out of 10 on a maturity scale in which 10 signifies “extremely mature.”
There are many reasons for this current state. For one thing, the concept of AIOps is fairly new. And as it has achieved “buzzword” status, there is no shortage of organizations and solution providers that have pivoted to recast their tracking, management and monitoring offerings as players in the category. This has contributed to confusion on both the buy and sell side of the equation.
The good news is that a consensus is emerging. Most operators are concluding that they stand to harvest major gains -- in both network performance and customer experience improvement -- if they can properly deploy AI across their operations. To do this, however, we will all have to reach a meaningful and shared understanding about what we mean by AIOps. In the process we will have to resolve some important Layer 8 issues -- the human factor -- before we can optimize the OSI stack to deliver the level of service across offerings that consumers already hope to receive and will soon come to demand.
If you look at the various formal definitions of AIOps and how the different analysts are characterizing the term, a couple of key attributes come up:
- The first is that it -- like DevOps -- is a silo-busting concept. It is meant to look across the various service delivery levels to provide insight on how activities taking place in one layer of the technology stack affect -- or create consequences for -- others.
- The second attribute revolves around the dynamic interaction of analysis and automation. The objective of AIOps is to create and deploy algorithms that evolve. They are supposed to learn and adapt in response to changes that are constantly taking place in today’s increasingly complex infrastructure environment.
This approach to defining AIOps puts immediate pressure on the organizational structures of most cable operators currently in the market. Different departments responsible for optimizing discrete segments of the infrastructure are happy to report that they are already making significant investments in AI to enhance security or improve network management or track endpoints and other assets.
The problem with this approach is that most of the analysis and automation taking place rarely leaves the silo in which it has been deployed. This makes it very difficult to develop an integrated perspective about overall operational performance. More importantly, it is nearly impossible to understand how changes to the infrastructure affect the customer experience.
This is a challenge that does not have a built-in champion within most cable organizations. Most network managers are reluctant to use a tool that is not native to -- or optimized for -- their layer. It is not difficult for them to characterize AIOps as an IT construct that they do not implicitly trust to handle their special circumstances. Ironically, the different IT disciplines -- and especially security professionals -- are likely to have an issue with AIOps for the same exact reason.
Without a serious organizational shift, this can be a challenging logjam to break through. The good news is that a growing number of cable operators are creating C-level positions -- such as chief customer officers, chief digital officers and others -- that do have a stake in getting a comprehensive perspective on how constantly evolving digital operations affect the customer experience.
Customer Centricity Driving AIOps Adoption
This shift is taking place as the cable sector comes to grips with how to survive in a hypercompetitive environment that has already altered consumer behavior. As certain doors close -- such as the ultra-profitable “walled garden” of the proprietary set-top box -- cable executives are looking to open new windows of opportunity by adapting their investment in network infrastructure in new ways.
From this perspective, there are two main drivers: one is defensive in nature; the other is more about scoring points on offense.
Defensively, AIOps offers an unprecedented way to glean comprehensive insights by uncovering and addressing technical issues that have a ripple effect across the operations. It improves operational efficiency by identifying the source of problems, identifying what systems are affected by those problems, and providing guidance that accelerates the resolution of the problem. In short, AIOps can offer competitive advantage by making the incident response process more efficient.
The more exciting opportunities, however, come from exponentially elevating the customer experience. AIOps can directly contribute to speeding up the development, refinement and delivery of new value-added services. It can allow operators to roll out upgrades more quickly and with fewer glitches by streamlining how new features and functionality are tested. In short, it is increasingly being recognized as a potential driver of significantly more average revenue per user (ARPU).
Open Transparency Key to Accelerated Adoption
Over the long-term, AIOps has all the hallmarks of being a concept that will be an important part of the cable industry’s future. The question is: When will it be comprehensively deployed?
The answer hinges on several factors. Those cable operators that have already made the decision to move forward with AIOps are now wrestling with whether to build or buy or to figure out an optimal combination of these options. Among the barriers to “buying” are concerns about lack of transparency.
Key constituencies within cable operators already have trust issues based on their current understanding -- or lack thereof -- about what third-party AI will deliver to their piece of the puzzle. This disposition may tempt them to pursue a “build” strategy, which may get them where they need to be eventually. In the process, however, they risk reinventing wheels that are already available from the technology partner community.
A more efficient alternative -- at least for the next year or so -- may lie in a hybrid approach of internal development supplemented by external solutions. For this to work, however, the AIOps tools will have to be open, transparent and modular.
The industry today has a wide array of offerings available that span the spectrum with opaque blackbox solutions on one end and open and transparent platforms on the other. I believe the latter will prevail in the market.
Blackbox offerings that spit out an answer and suggest a course of action without context or explanation for the analytical process are unlikely to find traction with technology managers who need to understand what is going on in their environment. The better way forward is to adhere to the direction we all received from our Algebra I teachers back in the day: show your work.
Transparent AIOps platforms that leverage the excellent work that has been done by various open source communities will address the trust issues that we must admit exist today. It will also do far more to contribute to the rapid education of key constituencies within the digital operations teams.
Chris Menier is general manager, digital transformation at Vitria.