Preparing For Artificial Intelligence Enabled IT Services: Method-Induced Noise

Vice President—technology and innovation at SLK software, an automation company with three decades of experience in IT transformation.

In my last article, I discussed machine-induced noise, and in the article before that, we discussed the requirement for eliminating noise necessary to embark on AI- and ML-based transformation.

Now let us look into method-induced noise and the elimination of that noise.

We define method in IT services as that which is a process or workflow. This workflow might be due to the requirement of a machine health or service management that provides support to that machine. For example, server support consists of several tasks that would require following a process—these processes would follow a workflow as per the IT service management system configuration, based on whether this is service desk or data center operations (level 1 to level n support).

‍

Method-Induced Noise

Unlike machine-induced noise, method-induced noise is "man-made." IT support teams have created processes and workflows with the genuine intention of making things simple and easily manageable. With monitoring, the intention was to prevent an error. In the process, a flood of events can happen that lead to threshold alerts which produce false alerts. This requires the suppression of these false flags, and the intention of prevention turns into a reactive, labor-intensive set of processes and workflows.

Usually, a service management system then creates a set of workflows to handle the generated ticket, describing what level of service is required and who should be assigned to the issue. These processes are impacting service levels and the end user experience.

‍

In recent times, a lot of digital transformation initiatives have been driven by CXOs in the form of RPA, chatbots and several other automation initiatives. Every ITSM product and monitoring product in the market today claims that it has enabled AI/ML chatbots and robotic process automation features, encouraging organizations to upgrade their newer versions. Monitoring was introduced to help—if that is not needed, one has to find a way to eliminate the ticket it is creating and not automate it away.

Whatever can be configured for monitoring can also be configured at the same level to deal with it in a simple, elegant way. If it was unnecessary to immediately take action, then it need not be monitored. This straightforward rule can help reduce this noise.

The above example provides a design principle for eliminating noise induced by machine-generated tickets. Another key place to look into induced noise is the service configuration.

‍

‍

“In recent times, a lot of digital transformation initiatives have been driven by CXOs in the form of RPA, chatbots and several other automation initiatives.”

Service Configuration Generated Noise

Whether it is a monitoring alert, a job failure or a user having an issue, an incident ticket is created. Then, the ticket follows an ITSM-defined life cycle: It first gets allocated to level 1, or the service desk, and then gets escalated to level n until it is resolved by assignment groups. Over a period of time, these assignment groups grow into hundreds in organizations as needed. They can include computation, storage, network, application platform, customer development platforms and middleware, etc.

Similar to the monitoring case where the inefficiencies are leading to unnecessary initiatives in cases like service desk voice-based systems and multichannel support (using chat, email, web and more recently voice-assistance) and unnecessary automation. AI/ML-based solutions are being introduced again as ITSM products in the form of intelligent routing and automated assignments, yet ultimately lead to more noise.

The fact of the matter is that if the requirement of reactive support for the user is being eliminated, then naturally, all these digital transformation initiatives would not be needed. If the hundreds of assignment groups can be converted into a smaller set of business processes aligned, self-sufficient groups, then the requirement of ticket routing can be reduced.

Reduce Bloat, Reduce Noise

Just like it is important to identify and eliminate machine-induced noise, it is integral to identify and eliminate method-induced noise. The good news is that if proper care is taken when eliminating the machine-induced, it can eliminate some method-induced noise that is directly linked to the machine. Identify the method-induced noise through two simple buckets: one that is machine-monitoring driven and the other that is service-configuration induced.

(Forbes - https://www.forbes.com/sites/forbestechcouncil/2023/01/03/preparing-for-artificial-intelligence-enabled-it-services-method-induced-noise/?sh=15f68a995bb4)

‍