Leveraging AI to Provide Better IT Service Management Using COBIT

I started my own Watson Assistant project last year. This was after I was informed of how successful IBM’s help desk, which is based on Watson Assistant, was. As an assurance specialist, I naturally decided to work on something along the same lines. Unfortunately, I never did complete my project – not because of technical issues but rather due to the simple fact that technology never supports a process that is not well-articulated nor fully documented.

As governments around the world grappled with the ongoing COVID-19 pandemic and announced various guidelines on distancing in public, announcements were made about working from home. This forced me back to the drawing board, to revisit my project, which essentially related to creating service chatbots on one of the assurance systems I work with.

IBM Watson Assistant, however, is not the only available technology supporting automated IT service management. Microsoft Bot Framework also provides the ability to build chatbots that IT departments can use and publish through Microsoft Azure. I experimented with Watson Assistant because of its cloud-based solution, which worked well as a test platform on which to build a prototype. It was more of a “plug and play” environment and a very good starting point in learning more on artificial intelligence. Whichever platform one explores, there is a need to have a systematic process to plan, build and run one’s project.

The pandemic has shown the need for organizations to review their processes and, where appropriate, automate IT services to build the capacity of the organization to continue providing quality services to their clientele. Users are now, more than ever, working at different times depending on how best they balance work and family responsibilities. The possibility of IT departments failing to resolve emerging issues within agreed service agreements has now increased. As such, the adoption of automated IT service desks is more feasible than ever.

In revisiting my project, I identified the following key/focal point areas, which I fine-tuned through use of COBIT 2019. There are some of my learnings:

1. Clearly define your goals for the project and articulate expected outcome. COBIT Management objective AP008 “Management relationships” outlines the following management practices that, if defined in an organization, can guide one to measure whether an IT unit is currently managing user expectation. These include:
   • AP008.01 Understand Business Expectations
   • AP008.02 Align IT Strategy with Business Expectation
   • AP008.05 Provide Input to The Continual Improvement of Service
   Ultimately, an analysis of relationship with the business should satisfy that IT is a value-add partner for the business.
2. Identify services and tasks that can be augmented using AI. For example, your first line of service desk can be managed through the use of chatbots. By utilizing Management Objective AP009 “Manage Service Level Agreement,” one can carry out a deep dive into the types of services provided and analyze whether ICT is able to meet service level agreements and how services can be revamped. Benchmarking against Management Practice AP009.04 “Monitor And Report Service Level” can provide insights on the capability of IT to provide timely service to users.
3. Review your current knowledge management within your IT service management. This is where a lot of work is required as most IT units have help desk systems. How data is collected and stored will be critical in building knowledge on user service requests. With most users working from home, most of the first line of assistance will involve diagnosis of reported problems and provision of guidance on how to resolve them. With the IBM assistant, one needs to create intents that are similar statements that require the same resolution. With a well-documented help desk system, this should be easy, as users normally word service requests differently but mean the same thing. In my project, this is where I failed, as I tried to develop intents without thoroughly analyzing the current data available. Leveraging COBIT Management Object “Management Service Request and Incident,” one can carry out an analysis of possible gaps in management of service requests. The management practice outlines provide a full cycle as follows:
   • DSS02.01: Define classification schemes for incidents and service requests
   • DSS02.02: Record, classify and prioritize requests and incidents
   • DSS02.03: Verify, approve and fulfill service requests
   • DSS02.04: Investigate, diagnose and allocate incidents
   • DSS02.05: Resolve and recover from incidents
   Where gaps have been identified, management can put on workshops within the ICT unit to build and refine available knowledge. Most of the work will have to be done at this stage. Although it seems simple, misinterpretation of user service requests will lead to incorrect responses assigned to service requests. Mapping of statements (user service requests) to service resolutions is a key output. Therefore, ICT management needs to have a formal process in place for defining and approving these statements. For better governance at this stage, a sample of users can be engaged to get clarity on how users best express their requests. Statements should capture users’ language and avoid use of IT textbook jargon.
4. Get an understanding of your organizational culture. Terms used within an organization are built through the organization’s culture and way of doing things. It is therefore important to take into consideration one’s organizational culture when designing a service. One must ensure that the solution provided to users is both workable and user-friendly. This will go a long way in building the system. This is not to say, though, that there is no room for customization, as there are already a number of standard user cases that can be customized to suit an organization’s needs.
5. Phase your project. As an example, the first phase of the project can focus on operational issues such as responding to user service requests; the second phase can go into predictive analysis to improve service delivery. Other phases can be added as the organization becomes more comfortable with the AI. Phased or modular project management approaches seem to have higher implementation success rates.

Conclusion
It is without a doubt that we will see greater adoption of AI to enhance IT service management. There has been keen interest of organizations in the retail industry in adoption of chatbots. This interest is now shifting to and will affect both public and private organizations because all users require quality service.