Whether it’s voice-powered personal assistants or connected smart homes, Artificial Intelligence (AI) and machine learning (ML) technologies are increasingly finding their way into our personal lives.
In fact, according to Gartner, “By 2020, the average person will have more conversations with bots than with their spouse.”
The rise of AI and ML is fueling conversations within the ITSM community about how to realize the benefits of these technologies in the workplace. However, knowing where and how to start can be challenging for even the most tech savvy IT pro.
Start with these questions when beginning your AI for ITSM journey:
How is the AI system developed?
AI for ITSM is not plug-and-play. Ticket categories, assignment groups, terminologies and SLAs vary across organizations.
Static, rules-based systems are too restrictive and limited to be valuable. AI systems that are overly broad are less accurate, lack sufficient domain knowledge, and require significant customization. All of which drives up the system’s total cost of ownership and time to value.
Additionally, having tight integrations with the organization’s existing ITSM, HR, and Financial applications ensures the relevance and accuracy of the actions taken by the AI system. And, allowing end users to interact with the system from their collaboration platform of choice (i.e., Slack, Skype of Business etc.) accelerates the rate of adoption.
For these reasons, and more, it’s essential that a team of both AI and ITSM domain experts are involved early in the data and workflow modeling process to ensure the AI system achieves outcomes based on the unique needs of the organization.
Where do chatbots fit in?
Chatbots can be highly effective at providing employees with answers to common questions and allowing them to order goods and services, in real-time, with little to no human contact.
However, not all chatbots are created equal. Those that are highly scripted and restricted to rigid answers to questions asked in specific ways don’t reduce call volume.
The most effective chatbots (more appropriately referred to as virtual agents) leverage a finely tuned mix of AI/NLP technologies (i.e., ML, Distributed Compositional Semantics, Fuzzy Logic and Ontological Engineering ) to engage in more dynamic interactions. Virtual agents are purpose-built to understand IT nomenclature, allowing them to differentiate between words like “Oracle” the corporation and “Oracle” the prophecy and recognize that “Salesforce isn’t running” refers to a software application failing to load properly on a user’s device.
That said, even the most accurate virtual agents will occasionally need to escalate tickets to human agents. So it’s essential that they are intelligent enough to assign incidents to the appropriate level of IT staff and auto-assign categories and priority levels.
How do employees benefit?
Employees want fast, easy access to IT services and support 24/7 and across multiple device types and locations. According to HDI: “as a rule of thumb, if an end user spends more than 10 minutes in a self-service portal, they cost the organization more money than calling the service desk.
A key value of an AI for ITSM system is its ability to present smart suggestions and execute automated remediations that are specific to a user’s persona, issue type and location, enabling them to quickly resolve their own issues.
By offering employees the ability to provide feedback and rate the effectiveness of the AI system’s responses and actions, IT can gain valuable insights and the AI system will become more intelligent continuously.
How can we cultivate adoption and sustain usage?
It’s recommended that IT service organizations gain a high level of familiarity, comfort and trust with the AI solution before deploying it to the broader organization. This can be accomplished by IT using the AI system internally to suggest the correct action, recommend experts and auto-assign/escalate incidents to the appropriate support staff.
It’s also important that IT set appropriate expectations, making it clear to business users that the AI system may not always immediately provide the “right” answer but will often require a back-and-forth dialog to appropriately fulfil a request or remediate an issue.
Lastly, IT should continually measure and message the benefits of its AI services to the business, demonstrating how they are providing faster response times to inquiries and more first call resolutions than traditional support channels.