How artificial intelligence is transforming a vital area of business
Lessons, advice, and encouragement
Infrastructure and operations (I&O) is a vital area of business that is often ignored because it isn’t “sexy”. Any company with significant numbers of customers will probably also have a large I&O division. As such it is ripe for AI transformation.
Currently, I&O is undergoing an AI transformation. However, it is still years behind the curve compared to many industry sectors. There are several ways to apply AI for I&O including chatbots and predictive analytics. A recent research note from Gartner highlights that most I&O leaders are still in the “Zone of Learning”, exploring the capabilities of AI. Over the next 5 years, Gartner predicts that 40% of I&O teams will use AI-augmented automation. In this blog, we will look at some of the ways AI will transform I&O.
What is Infrastructure and Operations?
A vital function that is often overlooked
Infrastructure and operations functions include help desk, infrastructure management and application performance management. These vital functions are ideally suited for AI transformation.
Infrastructure and Operations is an umbrella term for all infrastructure, operations and helpdesk functions within a company. In many ways, it’s the ugly big sister of site reliability engineering. Key I&O tasks are:
- Help desk operations and customer support. This includes customer contact, ticket management, and escalation management.
- Infrastructure management. This includes direct server management, service availability and cost management.
- Application performance management. This includes resource planning, application monitoring, and demand modeling.
What is AI and how can it help?
AI transformation could see a revolution in I&O. It promises to deliver significant cost savings, improved productivity and efficiency. It could even help with business transformation across the company. AI can be broken down into 3 broad fields:
- Natural Language Processing. NLP is the ability of computers to understand the meaning within natural language. This is the basis for chatbots and services like Alexa and Siri.
- Machine Learning. ML uses a set of data to train a computer to identify certain features as being correct or incorrect. Progressively the computer can be trained to be more and more accurate.
- Machine Perception is the ability of a computer to understand abstract concepts. Most importantly this includes speech recognition and computer vision.
So, how might AI transform I&O?
Help desk and customer support
Every business with end customers needs some form of helpdesk. This may be as simple as handling requests for application help, or it may include full support and escalation. There are three significant ways AI transformation could improve help-desk functions.
Increasingly, we are seeing companies deploying chatbots to act as the first point of contact with customer support. At present, these chatbots tend to be rather dumb, simply returning relevant help or offering to escalate to a human. One example of this is Revolut, who recently became one of the first fintech Unicorns. However, if you look at their community pages you will quickly find critical reviews of “Rita” and her performance.
Another key way AI might help is in intelligent ticket management. Machine learning and NLP will allow your team to process tickets more efficiently. The system will be able to track and intelligently assign tickets, escalating where needed. Reinforcement learning can then be used to improve the performance of your chatbot.
The growth of the cloud, containerization and serverless architectures have placed a significant infrastructure management burden on many businesses. Efficiency and cost reduction have been the drivers behind the adoption of these technologies. But the price to pay has been increasing complexity. Often an application may require tens of different services, each running in their own container. AI transformation can help with this task in a couple of ways.
You can easily train an AI to spot surges in demand far quicker than any human could. This means you can use the AI to handle resource allocation, spinning up new instances as required. Importantly, as soon as demand dropped it could reallocate resources, saving money. Similarly, an AI could be trained to understand the patterns in demand. Most applications see some form of diurnal variation in demand, but often this is too complex for a human to reliably analyze. However, machine learning is extremely good at spotting patterns like these, allowing you to have a system that pre-empts any expected spike in demand.
Application performance management
Performance management is often closely linked to infrastructure management. But it also includes aspects like spotting when your database is struggling, predicting long-term demand and resource planning. All these are ripe for AI transformation. As already mentioned, machine learning, especially deep learning, is ideal for prediction based on spotting patterns in data.
Drivers for AI transformation
According to the Gartner report, only a small number of I&O leaders report that they currently use AI. But some 30-45% expect to be using it in the near future. One of the reasons for this slow take-up may be a worry that the process will be difficult and expensive. Another reason may be that I&O is traditionally quite a conservative area of business. This is because it is critical to the performance of most businesses. So, there is a reluctance to do things that might “rock the boat” or cause problems. Finally, AI transformation requires real input from developers, machine learning experts, and data scientists. Often businesses will be reluctant to invest these sorts of resources without being able to see a clear benefit. In other words, there may be something of the chicken and egg here.
However, Infrastructure and Operations is the ideal business area to drive AI transformation across your whole company. Firstly, I&O staff are experts who will be able to understand the complexities of AI as well as its limitations. Secondly, the benefits will be easy to measure since I&O tasks and performance are generally already monitored closely. Thirdly, it is one of the areas where the benefits will be most widely felt. Improving the performance of your infrastructure benefits your customers, improves your efficiency and, ultimately, affects the bottom line.
Arguing for more rapid AI transformation
Before you can convince bosses to cough up the money, you may need to make it clear what the benefits are. One of the first things to note is that the benefits from AI transformation will make themselves felt pretty rapidly. Let’s take the example of AI chatbots. There are numerous frameworks for AI chatbots, and you can even access these using the “as a service” delivery model. This means that the investment needed from your side is relatively small. But having made that investment, the return should be fairly immediate. This is because your chatbot will now filter out all the “time wasting” or simple queries being passed to your customer service agents. This, in turn, frees up their time to deal with real problems, making them more responsive and improving customer experience.
Increasingly, developers are instrumenting applications and backends to allow for a better understanding of how they are performing. However, your apps are generating so much data you quickly arrive in the realm of big data. At this stage, machine learning becomes one of the most powerful data analysis tools available. Furthermore, ML is a scalable solution to a problem that is only going to get more pronounced over time.
Functionize’s experience with AI transformation
From the start, Functionize has concentrated on applying AI to transform the business of application testing. As a result, we have become experts in the field of AI transformation. The important thing we have learned is that no one-size-fits-all AI solution will ever work. Across our portfolio, we use a huge range of AI techniques. These range from natural language processing for autonomous test script creation to image recognition for template recognition. In some cases, we have even found that traditional data analytics solutions are actually better than AI. Hence our autonomous canary testing approach uses the Akaike Information Criteria for identifying user journeys.
The key takeaway from our experience is that AI can be truly transformative for businesses. However, you shouldn’t just apply it blindly. It is important to work out exactly what you want AI to do for you, and to select a suitable tool for that task. As with us, you may actually find yourselves combining AI with more traditional methods.
AI transformation is a hot topic across all areas of business. Infrastructure and Operations is no exception. However, the rate of uptake of AI within I&O has been relatively slow. In this blog, you have seen how AI impacts this vital area of business and gave arguments for speeding up adoption.