What is more convenient than a software robot that is available 24 hours a day to answer questions? The promise of software robots on a helpdesk seems big: higher customer satisfaction at lower costs. Yet this future promise is not immediately feasible in practice when you look at the state of Artificial Intelligence and the way service desks are organized generally.

Consumers will become increasingly familiar with the user interface of robots in the coming years. This trend started with personal assistants on smartphones and will show up in more and more portals, apps and other software products. The interface of an avatar enables customers to ask their question in spoken or written language and ensures that technology is transparent and no longer standing in the way. We have known this phenomenon for some years as consumerization of technology and we will see it again with smart systems. What customers use and like at home they will also want to use at work as well. Software robots that understand intentions behind questions and can immediately execute requests is the ultimate in user-friendliness. That is what end users want and expect.

High expectations regarding the application of virtual assistants have been set in motion by a number of remarkable breakthroughs in artificial intelligence. In recent years, Amazon, Facebook, Google, IBM and Microsoft have created various innovations by tackling their gigantic data collections with a combination of graphics chips and neural networks. What started with recognizing cats in Youtube videos quickly grew into systems that defeated people with complex games such as Jeopardy, Go and poker. Artificial intelligence now allows software robots to communicate in natural language, robots can operate software and they are especially able to learn. In addition, this technology has been democratized and brought to the markets in an unprecedented scale and pace. Almost all of modern AI technology is accessible in the form of affordable cloud services or fairly complete DIY frameworks. This makes the barrier of entrance quite low for the CIO, but also increases the urgency to act.


Many executives see unprecedented opportunities for applying robotics in service management. Statements such as that robots are the helpdesk employees of the future can be heard quite regularly. The ultimate consequence of this glorification of technology is that the existing staff of service desks will be decimated. This is both an undesirable consequence and a badly substantiated idea at the same time. There are sufficient reasons to temper expectations, according to fact-based research from sourcing consultancy METRI group. First of all, general artificial intelligence does not exist and it will take a long time before this capability can be built into software robots. Man is a master to interpret a conversation well bases on clues and information from the context and to reveal the right underlying intentions. This is difficult for a robot. Only in a very specific setting and context a software robot can listen and understand well. In this respect it is essential to start with the right expectations and also consider to let human agents supervise this robot technology to augment their productivity.

Another reason to temper expectations is that a software robot needs a lot of data in order to be able to achieve satisfying and useful results in conversations and the subsequent execution of requests. At many service desks too few good datasets are available to train virtual assistants in such a way that the system can withstand the expectations of critical users. Software robots need a lot of service data and logs in order to be trained well.

Finally, another important consideration deals not so much about technology itself but more about the way service desks are organized. Many support organizations are organized according to a break / fix principle with strict service level agreements in order to control costs as much as possible. An essential feature of virtual assistants is that this technology is actually the next generation of a self-help portal. The end-user experience should be a central principle in this, so that as many users as possible serve themselves with the help of robots. We are talking about predictive prevention here not about cost optimization. CIOs should ask themselves whether their service desk is ready for this technology.

Service desks with a traditional setup require a substantial investment in knowledge management, a high-quality self-help portal and a smart back-office in which software robots carry out their own actions according to fixed scripts. In this way, robotics in end user management must be a central part of a broad policy to make service management smart. No hype at all, but changing the organization to fully reap the benefits of AI.

Background and best practices

A robot that answers questions 24/7 seems to be the ultimate goal in service management. In the whitepaper ‘Robotics of end user management’ METRI Research explains what CIOs can expect of software robots and what are critical best practices for adoption of AI in end user management. More information and download