Beyond RFPs and chatbots: How AI is optimizing IT infrastructure and operations

Today, most of the big news in artificial intelligence (AI) is around the development of new, cutting-edge generative AI (gen AI) applications. From helping students write papers, to filling out RFPs in seconds, to (poorly) preparing legal cases, its successes and failures have been well-documented.

But what about more prosaic tasks? Some organizations are experimenting with deploying AI to automate aspects of IT infrastructure and operations, allowing valuable human expertise to be dedicated elsewhere.

“Generative AI is core to how many modern enterprises build new digital products to make money,” says Richard Warrick, Global Research Lead at the IBM Institute for Business Value. “But what if the same technology could radically change the business processes needed to design, deploy, manage and observe those applications?”

From automating resource-intensive processes like data center provisioning and DevOps to replacing onsite security personnel, here’s how intelligent AI automation is transforming IT infrastructure and operations. 

When AI was first being explored for business purposes, its ability to automate repetitive tasks that had previously required human input was considered its most valuable application. With the rise of generative AI and its many capabilities, that notion now seems quaint; but while the nature of the tasks AI can automate may have changed, the fundamental value of automation itself hasn’t.

According to a recent study by IBV, 80% of executives are automating their IT networking operations over the next 3 years, while 76% are applying the same AI skills to IT operations over that period.

“AI-powered intelligent workflows and IT automation tools are helping business leaders find competitive advantages, in terms of performance, that were eluding them before,” says Warrick.

Ten years ago, AI was being used to execute simple, rules-based processes, a skill known as rules-based automation. Examples of early rules-based AI include the operation of robotic arms and factory assembly lines. While rules-based AI tools were effective in certain areas, they lacked the flexibility and scalability to be applied to broader business problems. Rules-based systems relied on sets of pre-defined rules, and as the tasks they were being asked to perform grew in complexity, so did the number of rules needed for the systems to function, eventually creating systems that weren’t scalable.

In the 1990s, scientists began developing computer programs that could “learn” how to draw conclusions in ways similar to the human brain by using large amounts of data. This branch of AI, known as machine learning (ML), allowed the technology to tackle more and more complex tasks, such as speech and handwriting recognition, complex game play, and even the ability to assist in medical diagnoses.

Deep learning, a subset of machine learning that gained popularity in the 2010s, took the level of complexity that AI systems could handle to whole new levels. Training on multilayered neural networks, deep learning programs simulated the complex and nuanced ways in which human brains made decisions, making it possible for AI to build applications, interpret images and video even respond to voice and text prompts the way humans do.

Today, thanks to ML and deep learning, AI automation has evolved from simple, rules-based processes to rich, sophisticated models trained on massive datasets that can perform many of the same tasks as their human counterparts. This new wave of AI tools—known as “intelligent automation”—is helping organizations improve their IT infrastructure and operations by streamlining business operations, analyzing data and resolving complex issues that previously required human attention.

Modern enterprises need their most brilliant, technology-focused minds concentrated on initiatives that have the potential to deliver the most business value, and right now, that means developing gen AI applications. According to another IBV study, 64% of CEOs reported feeling pressure from investors, creditors and lenders to speed the adoption of gen AI, and more than half said they felt the same pressure from their employees. 

But gen AI applications require the support of complex infrastructure to facilitate the collection, processing and secure storage of massive volumes of data. Previously, that responsibility fell to teams of IT managers, engineers and data scientists, but what if it could be accomplished by AI?

Intelligent AI automation taps into the power of specialized skills like computer vision, natural language processing (NLP) and other advanced AI technologies to solve highly complex business problems. AI models trained on massive datasets using machine learning (ML) and deep learning can analyze data from applications and systems, quickly identify patterns and adjust resources and processes accordingly before problems occur.

Computer vision is AI that interprets images and videos like the human brain. AI models use ML and deep learning to repeatedly analyze data, eventually identifying relevant differences in images and videos. For example, an AI model being trained to secure a home would need to be trained on thousands of hours of home security footage so it could learn to recognize a likely intruder.

In IT infrastructure, computer vision is used for a variety of tasks that previously required human input, including predictive maintenance, system monitoring, data stream processing and more:

• Predictive maintenance: Specially trained computer vision systems use image recognition algorithms to detect problems in physical components like cables and servers before they result in network downtime.

• System monitoring: Computer vision can accurately analyze large amounts of data from a variety of diverse sources much faster than a person can. Examining camera angles of subway tunnels, freeways and buildings would take a human being hours, but a computer vision system can do it in real-time.

• Data stream processing: Computer vision systems process and analyze enormous quantities of data delivered from physical sensors that track key metrics like temperature, humidity, air speed and more. Relying on computer vision to detect sudden changes in conditions often alerts organizations to a problem immediately.

Natural language processing (NLP) is a field of AI that focuses on how computers can be trained to understand and communicate using human language. NLP helps systems recognize and understand human speech and generate text in response to prompts.

Recently, NLP was critical in the development and launch of ChatGPT, a groundbreaking chatbot that can understand and generate human-like text in response to questions and prompts.

In IT infrastructure and operations, NLP helps organizations with a variety of tasks that previously required human input, such as improving user experience, ticket resolution and sentiment analysis:

• User experience: NLP enables users to ask questions about complex IT issues in the way they would speak to a customer service representative and receive helpful answers. Chatbots trained on vast amounts of technical knowledge and equipped with NLP skills can take the place of customer service representatives with years of acquired technical knowledge.

• Ticket resolution: AI NLP systems can analyze incoming tickets and accurately prioritize and categorize them by importance and type. AI systems can even be trained to resolve the issues themselves, request human intervention when needed, or take other appropriate action.

• Sentiment analysis: NLP systems can conduct sentiment analysis on user feedback, surveys and even social media posts, accurately gauging the emotions behind responses and identifying areas for improvement. Additionally, NLP systems can help organize specialized technical information, improving documentation and IT knowledge sharing across an organization.

Applying intelligent AI automation to IT infrastructure and operations is transforming how IT managers monitor and optimize their systems and allocate critical resources. Here are four examples of areas where the technology is helping transform processes, reduce costs and identify meaningful insights into core business practices.

AI is extremely adept at spotting patterns in data, making it a perfect fit for analyzing the massive amounts of data that flow through enterprise data centers every day. Data center operators have begun to embrace AI to help them spot patterns in data and identify opportunities for automation and process streamlining, a key part of increasing return on investment (ROI) for digital transformation initiatives.

One area where AI is already improving data center operations is in energy usage. AI systems can monitor and dynamically adjust cooling systems and manage power consumption helping enterprises save millions—in one case, lowering a data center’s energy bill by 40%.

AI is increasingly being used to automate aspects of data governance, the process of maintaining data integrity and security while it’s collected, stored and processed. With the rise of generative AI, businesses are discovering they need to collect and manage much more data than they have in the past. Since the data they need is often collected in one place and stored and processed in another, staying in compliance with the applicable compliance laws can be challenging. AI systems automate certain aspects of the compliance process, updating based on laws and regulations without human input, making the entire process more efficient and secure.

AI is playing an increasingly important role in observability, an aspect of IT operations that helps organizations understand the condition of complex systems based on those systems’ outputs. Observability can be applied to a variety of infrastructure components, including servers, applications, network devices and more.

AI models trained for observability purposes monitor data from these systems and analyze it for errors and inefficiencies. Using advanced intelligent AI automation, some AI systems can even pinpoint the root causes of certain issues and take appropriate actions before they impact the availability, performance or security of applications.

In addition to monitoring system and application performance and availability, AI is also transforming provisioning, the process of making hardware and software resources available to systems and users. Today, advanced AI systems automate provisioning, helping enterprises allocate cloud computing resources more efficiently so machines don’t sit idle and overall performance doesn’t drop. The market opportunity for intelligent AI automation in the provisioning process is significant: According to an industry report by Flexera, more than 32% of cloud spending is wasted on poor provisioning.¹

AI systems are being used to improve DevOps, a method of software development that bridges the gap between coders and IT operations. Some enterprises have used AI to automate software testing, leading to faster development. Others have deployed AI algorithms to analyze pipeline data and improve resource allocation. Still other enterprises are increasingly relying on generative AI to write code, test it, identify bugs and even suggest potential fixes.

According to IBM Fellow Kyle Brown, AI isn’t just being used to automate certain aspects of DevOps, but entire platforms. “Today, you can implement a common AI DevOps platform that is completely configuration-driven and automated,” he says. “No matter what a dev team is working on, if they’re building it on one of these platforms, they will be in compliance with guidelines the enterprise has set.” 

While generative AI and its potential for business applications may still garner most of headlines, organizations applying AI to the systems and processes that underpin IT are discovering new ways to reduce costs and transform out-of-date systems and processes. From automating resource intensive tasks like provisioning, compliance and software testing, to monitoring complex systems for intrusions and scouring massive datasets for real-time for insights, the potential for innovation in the space is limitless.

Modern AI IT infrastructure and operations solutions (AIOps solutions) deliver a complete, fully-integrated set of AI-powered tools that automate processes and deliver powerful insights into the performance and health of systems and applications. “These modern tools are a godsend for IT operations teams,” says Brown. “Take just one area— planning, for example—and we’ve seen intelligent AI automation cut planned spending on hardware and additional resources in half.”