The Need for Speed

There was once a day when user experience experts were focused on visual and interaction design. In other words, they were focused on making an app easy and pleasant to use. With the advent of mobile consumer apps, that focus has evolved to not just include visual and interaction design, but performance as well.

With the introduction of artificial intelligence (AI), the consumer world has redesigned its services to address the new performance requirements for conversation-level speed, but enterprise software vendors are quite behind. And, that’s an issue. Because without speed, employees will never adopt applications that could significantly increase productivity and reduce costs for the enterprise.

A study conducted in 2017 by Kony found that 87% of banking and finance leaders reported abandoning a mobile app due to poor performance.
Another study conducted by AppDynamics in 2017 found that 80% of users will delete an app that has poor performance.

Why is enterprise software behind when it comes to speed? Unlike the consumer world that has built AI apps on new technology, the enterprise world has tried to bolt it on to an architecture that was never designed for it – and experience tells us that is never the right answer. That is especially true in the world of IT Service Management (ITSM) where performance issues have been known to occur, even when low employee adoption rates of 10 to 15% are already the norm. Employees require AI based apps to respond with speed – all of the time.

When the World Was Simpler

ITSM platforms were designed well over 10 years ago when technology was dramatically different. The focus back then was on virtualization of physical hardware, and as ITSM vendors moved to the cloud, they took advantage of this trend by virtualizing their applications as well. This helped these vendors scale global data centers efficiently and cost-effectively – to a point.

Each virtual machine (VM) was optimized for specific customers based on size and usage. If customers ran into performance problems related to VM resources, they needed to pay for an upgraded configuration – which sometimes meant the vendor needed to put the VM on a larger or a dedicated computer.

The other issue was resiliency due to the primary/secondary data center architecture. If a customer’s primary VM went offline, the vendor would reconfigure the customer’s traffic to the secondary VM. This switchover could take minutes, during which time the service was unavailable, and work could be lost.

In all fairness, speed wasn’t as important when ITSM users consisted primarily of help desk agents manually entering tickets in a form. But the world has changed.

AI Has Changed Everything

AI has become pervasive in the consumer world, and enterprises are now beginning to embrace AI to improve their employee self-service experiences while reducing costs. With AI, the need for speed is changing everything.

Our research shows that employees expect a response from a Virtual Support Agent (VSA) in less than 3 seconds – and that includes network latency. A typical VSA conversation involves a multitude of AI technologies such as natural language processing (NLP), deep learning, neural networks, spell check, real-time translation, paraphrasing, and more. These technologies are CPU and memory intensive and can only be delivered with speed if the service was initially designed to scale with this new world in mind. And don’t forget that responsiveness needs to be the same whether there are hundreds or thousands of users simultaneously using the app.

Legacy architectures just were not built to meet the demands of this new world.

Elastic Architecture is the New Requirement for the Enterprise

Companies like Google, Facebook, Amazon and others have learned the painful lesson of needing to design their services to scale with demand versus bolting onto older architectures. They understand the need for speed and have leveraged the best in elastic architectures to meet consumer expectations. The trick is achieving this in an enterprise environment, while keeping an individual customer’s environment secure.

So, what is an elastic architecture for the enterprise? Imagine an architecture where a customer’s workload is run inside a container. The container is not tied to any specific machine, and in fact can run in any data center. A container is designed to handle a specific number of employee transactions per second. Understanding the complexity and high cost of AI, it can predict when a container will reach capacity and add containers as required to ensure performance goals are met.

Of course, the limiting factor here is data center resources. However, by leveraging AWS’ massive footprint, a service has no boundaries. Load balancing is then used to ensure proper distribution of traffic across all containers for a given customer. The resulting architecture can therefore scale to the largest of organizations.

What Good is Speed if the App is Down?

The days of long term SSL connections to an application are over. In today’s world, it is all about using stateless connections to an application using REST APIs. With stateless connections, a given employee’s request can be sent to one container, while the next request is sent to a different container. There is no state maintained across individual requests, which means no long-term relationship between an endpoint and a container. If a container fails, traffic is immediately redirected to an alternate container – and employees won’t even notice.

Combining the elastic and stateless endpoint architectures provides a high degree of resiliency. Combine this with a solution that takes advantage of multiple data center availability domains, and you end up with the optimal resilient architecture. This means that even if a whole data center were to fail, the service would immediately be spawned in another data center and end users would not be impacted.

When AI Technology is Built on New Architecture

Trying to build an AI based app on older architecture will never deliver the performance and resiliency that’s required by chatbots and virtual agents. An AI based app built on elastic and stateless architectures, on the other hand, will deliver performance and resiliency across the entire stack, and can scale to meet the needs of the largest customers.

At Espressive, we designed for today’s world and built our solution on elastic and stateless architectures. As a result, Espressive customers get the speed and resiliency that they need to enhance the overall employee experience – even when used concurrently by hundreds or thousands of employees.

If you are considering using AI to deliver an exceptional ITSM employee experience, let me know. I’d love to show you how you can transform the employee self-service experience while leveraging your current ITSM investments.