On the Way to Invisible IT Support

Jayant, before we talk about MyIT: What is your role at Siemens?

Jayant Deulgaonkar: I work at Siemens AG in the area of Service Integration and Management, or SIAM for short. Our team is set up globally and focuses on better integrating IT services with one another. The aim is to orchestrate services in such a way that they work as smoothly as possible for users. This includes topics such as process governance, consulting for services, IT asset management, management reporting and a large part around the ServiceNow platform. I am globally responsible for this platform.

What is MyIT and what is the idea behind it?

MyIT is the central entry point for IT services at Siemens. The idea was born around 14 years ago because, at that time, there were many different portals and applications. For employees, it was often unclear where they had to go with which request. For one topic there was one portal, for another topic there was another. MyIT was intended to solve exactly this problem and create one central point of contact.

The principle was clear from the very beginning: one door for everything that users need in an IT context. This includes content, navigation, search, knowledge articles, news, information about their own devices, licenses or services, as well as the possibility to report incidents, request changes or order new devices.

That sounds like an early example of employee experience in a large enterprise.

Yes, absolutely. The term may not have been as present back then as it is today, but the basic idea was exactly that. It was about making the IT experience for employees simpler and more centralized. Our MyIT platform was rolled out globally. Multilingualism was also important from the start. In the past, translations were maintained manually. Today, this is technologically very different, for example with live translation.

How has your platform evolved over the years?

In the beginning, the focus was strongly on bringing services onto this platform and showing them the benefits of workflows. That was the first phase. Many services previously had a great deal of freedom in their own solutions. They first had to understand that, although they would give up a certain degree of this freedom, they would benefit from shared processes and a better user experience.

In the second phase, acceleration came in. More and more services wanted to move onto the MyIT platform. A lot was developed in-house, in some cases even individually. Over time, however, this created a very high level of technical complexity. After around ten years, our own platform had grown significantly. It worked, but technically it was no longer as close to the standard as it should have been.

What did you do about that?

We rebuilt the platform almost completely and brought it back closer to the standard of the ServiceNow platform. That was a two-year journey. The goal was to renew the technological foundation, standardize the platform and, at the same time, not simply lose the existing automations. In January 2025, we were fully live. For a platform of this size, it was remarkable that there were only around six hours of downtime for users.

Why was this standardization so important?

Because it forms the foundation for everything that comes next. Once the MyIT platform had been rebuilt and was once again stable and closer to the standard of the actual ServiceNow platform, AI functionalities also came more strongly into play. The timing was very good. We were prepared when the new possibilities became available.

How did you then get started with the use of AI?

We participated in ServiceNow’s AI Lighthouse Program. At the beginning, we tried to identify classic problem areas and find suitable AI solutions for them. But that took too long. Then we reversed the approach and asked: Which solutions does this platform already offer, and how can we use them in a meaningful way? That brought significantly more speed.

In the first six months, we had implemented only two AI skills, which were also not used very much. After changing the approach, we were able to roll out 17 skills that were available more quickly and were used much more actively.

What exactly do you mean by skills?

These are functions that support employees when handling incidents or requests. One example is the automatic creation of knowledge articles. Often, a great deal of knowledge is contained in a complex incident. Several people work on it, there are many notes, and in the end the solution is documented. But if no knowledge article is created from it, this knowledge is lost.

With one skill, a formatted knowledge article can be created from such an incident with one click. The responsible person only needs to review it, adjust it and publish it. Something that might previously have taken a day can now be prepared in just a few seconds.

What is the effect beyond pure time savings?

The real added value is not only that someone has less effort. If knowledge articles can be created more quickly, more of them are created as well. And good knowledge articles help to avoid incidents. Users find their answer directly in the portal or via a bot and do not need to open a ticket at all. This is a qualitative effect that is often not visible enough in traditional ROI calculations.

What role does data quality play in this context?

A very large one. Every automation needs a solid data foundation. This is not only true for AI. If data is only partially correct or inconsistently structured, automation remains limited as well.

Over the years, many fields and attributes had been created in our system. In some cases, different fields were used for the same information. A simple example is a serial number. If there is one standardized field for this, automation can work with it well. But if there are 15 different fields in different tables, it becomes more complicated. It is not impossible, but it is worse for performance, comprehensibility and user experience.

How did you deal with that?

We rationalized significantly. At one point, an incident form had 365 attributes. We analyzed which fields were really necessary and reduced the number to just over 50. This made the data model clearer. Everyone understands the meaning of the attributes in the same way. That makes automation and the use of AI much easier.

Another central topic is governance. How do you deal with that?

That remains a challenge. When AI agents can make suggestions or perform tasks, the question arises as to who controls which solution is used and according to which rules the work is done. There could be a ServiceNow solution, a ChatGPT-based solution or another agent solution. If many systems want to solve the same problem, governance becomes critical.

In addition, there is the issue of hallucinations. AI generates content, and you can never rule out 100 percent that something is wrong. That is why you have to define what level of risk is acceptable and where human control remains necessary. This is not trivial.

Does that mean you do not yet allow agents to act autonomously?

Not where they make decisions or trigger actions. If an agent summarizes something, it does not necessarily require control. But if it performs an action, we rely on supervision. For example, if a system recognizes that certain steps could help with a PKI card, it may suggest clearing the cache. However, the user must actively confirm this action. We currently do not allow fully autonomous actions.

At the same time, there is a vision of touchless IT. How close are you to that?

This vision is still young. It was formulated in 2026 and is aimed at the year 2030. The goal is to make support as invisible and effortless as possible. Technology is developing very quickly, so we are optimistic. But it will take time, especially when it comes to governance, security and control.

What do you expect from ServiceNow’s AI Control Tower in this context?

I am particularly interested in the governance and security aspects. Especially when agents become more involved in operational tasks, an overarching control layer is needed. If what we have seen becomes reality, this could close an important gap. Not only for ServiceNow, but for the market as a whole.

What is the next big step for you personally?

For me, the topic of the AI Control Tower is central. Security, governance and control are the biggest open questions. If these topics are solved well, AI can have a much stronger impact in IT service. Then we will truly come closer to the vision of largely invisible, effortless support.