With AI’s transition from disruptor to dominance well underway, organizations around the world are investigating how best to implement this groundbreaking technology.

AI’s capacity to automate processes and provide unique and complex insights can help organizations work more efficiently and reduce costs through better decision-making, potentially giving them the edge they need to compete in the marketplace.

The application of AI to enhance and automate various aspects of IT operations and management within organizations is called AIOps – short for AI for IT operations.

AIOps combines data analytics and machine learning to analyze the massive amounts of data generated by an organization’s IT infrastructure and applications. The goal is to make IT operations more efficient and reliable by monitoring performance, automating routine tasks, detecting issues and flagging them for human attention, and providing actionable insights.

It can help IT teams streamline their operations, reduce downtime and lessen the impact of service disruptions by being more proactive in addressing issues before they affect users.

We asked two experts in the field for their insights into several aspects of AI-led innovation in the field of networking:

AI is enhancing the performance of networking platforms

When we asked our interviewees about how AI can improve the performance of a networking platform, this is what they had to say.

“AI can be used to enhance network platform performance through network optimization, predictive maintenance and intelligent monitoring. It can analyze network traffic patterns to identify potential bottlenecks, and it can use data from sensors and other sources to predict when network equipment may fail, allowing for proactive maintenance.

“Lastly, by analyzing network traffic, event and performance data in real time to identify anomalies, AI algorithms can alert operations teams to potential problems before they cause significant issues.” – Casey Kindiger, Grok

“In our experience, AI can significantly enhance the performance of networking platforms in multiple ways. For example, through leveraging AI, we can create intelligent algorithms that optimize data routing, reduce network congestion and improve overall reliability. Furthermore, AI-powered predictive analytics (supervised learning) can proactively identify potential network failures or bottlenecks, allowing for preventative maintenance and ensuring seamless service continuity.” – Iain Brown, SAS

So, with the help of AI, network management can become more intuitive and responsive than we’ve ever imagined. Removing mundane tasks from employees’ duties gives them time to focus on other, more important projects.

Within the parameters provided by IT teams, AI can find more efficient ways to use resources, and identify and help eliminate suspicious activity before it can cause any real damage. Reduced costs, better security and reduced latency can be delivered simultaneously with the use of AIOps.

AI-enabled network improvements

Network improvements led by AI-enabled programs have far-reaching benefits that can affect the entire system.

Network performance monitoring: One of the challenges of large IT operations is ensuring a seamless network experience. With AI algorithms, vast swaths of data – network traffic, operational data and performance metrics – can be analyzed in real time. Potential problems can then be flagged for review before they become more serious.

Network automation: AI can take on a range of tedious tasks in IT operations, such as the configuring of devices. Removing these tasks from humans leads to fewer errors while saving time and increasing efficiency.

Network security: IT teams have to protect their valuable data from security threats such as data thieves or malicious software. AI helps to identify potential threats by running the security-event data through its program to detect anomalies and catch suspicious activity before the integrity of the system is compromised.

Network optimization: This can mean the difference between coming out ahead of your competition or being left in the dust. AI algorithms are well equipped to optimize network resources, including bandwidth and latency, thereby improving the user experience and network performance.

Network troubleshooting: In IT departments, problems were traditionally solved by team members getting together to troubleshoot the issue and brainstorm solutions. AI does this better, faster and more effectively, saving time and money. AI-enabled programs can even take corrective action automatically and anticipate potential problems.

Our experts also explained where they believe AI will make the most profound impact:

“AI is readily applied to IT operational data to deliver predictive maintenance and intelligent monitoring and automation. Predictive maintenance solutions reduce operational costs by avoiding incidents and costly problem investigations that happen after the fact. Using AI to monitor network performance in real time allows operations teams to quickly identify and respond to issues.” – Casey Kindiger, Grok

“AI-enabled solutions in networking predominantly focus on predictive maintenance, anomaly detection and network optimization. Predictive maintenance algorithms can anticipate and mitigate network failures before they occur, leading to less downtime and better service. Anomaly detection helps identify unusual network behavior that could indicate potential threats, leading to improved security. Network optimization can benefit from AI to adjust traffic flow and resource allocation dynamically, leading to more efficient network utilization and better performance.” – Iain Brown, SAS


There are two main approaches to implementing AIOps in organizations: doing it yourself (DIY) or working with a managed service provider (MSP).

When organizations go the DIY route, they take responsibility for procuring, deploying and managing all components related to their AI systems.

This approach has lower upfront costs but requires significant effort from internal IT staff or third parties with expertise in the specific AI technology being used. Additionally, organizations must budget for the long-term costs associated with this approach.

Alternatively, an MSP can provide a comprehensive package of end-to-end services that include hardware, software, training and support.

This model offers more flexibility due to the MSP’s experience and specialized resources. There would also be lower upfront costs for organizations that are implementing AIOps in this way.

The challenges of implementing an AIOps platform in-house

Our experts didn’t shy away from stressing the time and effort it takes to implement a proper AIOps platform.

“Building and implementing an AIOps platform in-house can be a complex process, both in terms of cost and time. For example, a key challenge is integrating data from different sources, such as logs, metrics and events, into one repository. This requires significant effort in data normalization, enrichment and transformation.

“Choosing the right machine-learning algorithms and models to analyze and correlate data is another challenge. You need a deep understanding of the data and an ability to identify patterns that can be used to develop predictive models. Once the algorithms have been selected, training the models needs significant computing resources and data-science expertise. This process can take several iterations to achieve an acceptable level of accuracy and reliability.” – Casey Kindiger, Grok

“Building and implementing an AIOps platform in-house can be a considerable undertaking, in terms of both cost and time. From a cost perspective, you need to consider not only the technology stack and infrastructure you need but also talent acquisition and retention, which can be substantial given the stiff competition for skilled AI professionals. The time required to train the models, validate their performance and ensure seamless integration with existing systems can also extend project timelines significantly.” – Iain Brown, SAS

So, implementing an AIOps platform in-house can be costly, time-consuming and complex. Organizations must have the resources and expertise to develop, deploy, maintain and update this technology.

Additionally, they should consider the cost of hardware, software licenses, storage capacity for data processing and analytics applications, as well as the personnel costs to maintain the system.

Furthermore, it can take significant time to build an AIOps platform tailored to an organization’s specific needs – including training datasets – and ensure the accuracy of its output. Organizations must carefully weigh the costs and benefits of doing this in-house.

What is the future of AIOps?

The potential for AIOps is tremendous, as it is expected to revolutionize how operations teams manage IT systems. However, deploying and maintaining an AI-powered system is no small feat.

Potential hurdles include data availability and quality, algorithm selection, model training and validation, deployment and integration with existing systems, and maintenance and support. Additionally, organizations have to ensure their AI models are designed ethically and securely to prevent bias or the misuse of data.

According to our experts:

“AIOps adoption will accelerate over the next five years until it is a pillar of network strategy as more organizations realize the benefits of using AI and machine learning to automate IT and network operations. Successful organizations will develop AIOps platforms that employ a variety of proven AI algorithms and techniques to improve accuracy. As AIOps algorithms and models become more sophisticated, they will be able to provide more accurate and reliable insights into IT and network operations, helping organizations to identify and address issues more quickly.

“AIOps will become more closely integrated with network engineering and DevOps as organizations look to optimize their entire software development and deployment lifecycle across increasingly software-centered network architectures.

“AIOps initiatives will also increasingly be used to improve business outcomes such as reducing downtime, improving the customer experience and increasing revenue. It is fundamentally a capability built on machine learning and AI algorithms – but how to architect a targeted solution and interpret the results can require a deeper understanding of AI than may be available in many organizations. Designing AIOps with built-in methods and techniques for ‘explainability’ can accelerate adoption by removing a significant potential hurdle.” – Casey Kindiger, Grok

“The future of AIOps looks promising, given its potential to revolutionize network management by enhancing efficiency, security and reliability. However, a successful rollout is not without challenges. For one, integrating AI capabilities into existing IT operations can be complex, especially in legacy systems.

“Also, the quality of AI outputs depends largely on the quality of input data. Poor data quality can lead to inaccurate predictions. Lastly, there's the human factor – despite the benefits, people may resist change due to fear or misunderstanding of AI. Proper training and change management will be crucial in overcoming this hurdle.” – Iain Brown, SAS

Looking ahead

Overall, the future of AIOps is bright. With advances in technology and increased investment in research and development by vendors in this space, it’s only a matter of time before we start seeing significant improvements in managing network performance, availability and cost.

With the right strategy, resources and expertise in place, organizations can make the most of AIOps solutions to overcome potential hurdles – such as data availability and quality, algorithm selection, model training and validation, deployment and integration with existing systems – and use AI to their advantage.


Read more about SPEKTRA, NTT’s next-generation platform for managed network services that uses advanced AIOps and automation.