Artificial intelligence (AI) is nothing new. Making computers think in the way that people think has been debated for as long as computer science has existed as a concept, and there have been several booms and busts in AI research over the last seventy years. However, more recently a shift has occurred, in which a more specialized notion of AI has begun to proliferate in often subtle, hidden ways, and in almost every part of our daily life.
AI is being used in the automotive industry to improve the driving experience, in the healthcare industry to diagnose patients and identify potential treatments, and in the agriculture industry to improve operational efficiency by automating essential farming processes. One development in which AI is relatively visible is in the evolution of the chatbot. As an intuitive customer experience, the chatbot has transformed in recent years: it has grown smarter and more humanlike in its responses, can provide contextually relevant information 24/7, and frees up human workers for more complex interpersonal interactions.
This emergence of AI as an operational component of businesses across industry verticals poses a specific challenge for IT operations teams. Feeding AI systems with the information they need to operate effectively, monitoring and processing the information which they produce, and overseeing increasingly complex user-machine interactions means handling data on a scale which itself demands automation. In short, it’s becoming clear that the age of AI will also be the age of AIOps.
Incorporating AIOps into the IT workflow is not, however, as easy as flipping a switch. For the integration to be a success, it is important to recognize your organisation’s key requirements and build a methodical approach on the basis of those requirements. Here are a few key considerations:
1: Consolidate data
As services and devices become increasingly smart across the board, Internet connectivity has become a general requirement. A chatbot, for instance, not only potentially gets its responses from a centralized system, but feeds usage analytics back in to inform development teams. With this comes the truth that data loads are getting larger, increasingly distributed across systems, and more diverse in type and form. Due to this, step one is for teams to understand which data are a priority and build a platform that can collect, analyse, and store those data. By aggregating information from a variety of cloud sources (traditional, private, public, and multi-cloud) into a single data lake, the quality of data is improved and a strong foundation for AIOps integration can be established.
2: Anticipate diversity
In the relatively simple case of a chatbot, the diversity of data coming into a system might include structured analytics data such as dwell time and exit surveys, unstructured interaction data such as human-produced text, and even rich media data such as voice input. Monitoring multiple types of data concurrently can generate invaluable insight on how a business’s services are performing. It is therefore vital that IT teams not only collect this information, but make it available in a way that makes correlations between the data identifiable. By doing so, an AIOps solution will be able to predict, identify and develop prescriptive recommended solutions to issues in real time through automation.
3: Monitor everything
Given that different systems have their own native monitoring tools and IT professionals have varying personal preferences around which monitoring tool is best, many teams are now using twenty or more of these tools at a time. While integration limitations are also a factor, this profusion leads to lost time as responsibility is subdivided across parts of the team. As the introduction of AIOps automates parts of the operations strategy, success depends on bringing these various monitoring activities into a single view, increasing productivity by aligning workflows with business priorities.
Reaping the benefits
A well-chosen and well-integrated AIOps solution can bring benefits which should entice any IT team, regardless of the degree to which AI is already at play in the business. Whether it’s quicker identification of advanced threats, better correlation between change and performance, improved efficiencies within IT, or faster time to deliver, the visualization and automation enabled by AIOps leads to better alignment between IT services and business outcomes. As the result is an improved employee and customer experience, it’s no surprise that AIOps projects are growing globally.
More in AI & Machine Learning
Conference Session Preparing the data center network for the downstream effects of 5G and edge