How AI Is Transforming the Way We Interact with Enterprise Software

Anders Landig:
When you look at the market, there seems to be a widening gap. On one side are companies already working extensively with AI and developing their own solutions. On the other are organisations still at the very beginning, testing their first use cases. Are you seeing the same thing?

Jonas Degener:
Yes, we definitely see that divide. Interestingly, however, the difference rarely lies in whether a company wants to use AI or not. The decisive factor is the foundation.

Many organisations initially focus on the AI model itself. In reality, however, it starts with data, processes, structures and workflows. The AI model is often only the final layer. The most successful projects are those where companies first clean up and optimise their data and process landscape before building AI on top of it.

We often see companies wanting to build an agent or chatbot as quickly as possible. If the foundations are missing, success usually remains limited. If those foundations are put in place properly, meaningful and sustainable results are often achieved far more quickly.

Anders Landig:
Why do so many companies struggle with that step?

Mirko Pleli:
Because expectations are often shaped by personal experiences with AI. Today, anyone can try ChatGPT, Claude or Gemini and quickly experience impressive results.

In an enterprise environment, however, different rules apply. Data security, governance, compliance and system landscapes play a central role. It cannot be acceptable to transfer company information freely to any model.

That is exactly why the groundwork is so important. Data must be structured, processes must be defined and a secure framework must be established. Only then can organisations achieve the same positive experiences many users are familiar with from their private lives.

Jonas Degener:
There is another important factor as well: in business, it is not enough for an AI system to deliver good results occasionally.

If a personal assistant gives a wrong answer once, you simply try again. But if AI correctly classifies only seven out of ten service tickets, it is quickly perceived as a failure.

That is why we actually use surprisingly little AI in many projects. Our principle is: as little AI as possible, as much AI as necessary. Stable processes and a clean data foundation are what truly matter.

Anders Landig:
Beyond the foundations, many companies also seem overwhelmed by the sheer number of technologies available. Are you seeing that as well?

Mirko Pleli:
That was particularly evident over the past two years. The market was flooded with new solutions, platforms and promises. Many organisations simply did not know which path to follow.

The situation is gradually becoming clearer, though. Providers such as SAP are investing heavily in secure platforms, agent frameworks and integrated development environments. As a result, companies now have clearer points of orientation. The range of options remains vast, but the path has become easier to understand.

Jonas Degener:
Overwhelm is often caused not by a single tool but by the sheer number of different tools being introduced simultaneously.

Every employee experiments with something different, and every department follows its own approach. That is why we recommend first defining a common technological framework. In many cases, this consists of hybrid architectures that combine SAP and Microsoft technologies.

The goal is not to restrict innovation but to create a shared toolbox. Otherwise, organisations quickly end up with chaos rather than added value.

Anders Landig:
Which developments do you currently consider particularly significant?

Jonas Degener:
The most exciting innovations are currently happening less within the AI models themselves and more within the data platforms underneath them.

Major technology providers are investing heavily in modern data architectures such as Microsoft Fabric, Databricks and SAP Datasphere. These form the foundation that makes productive AI adoption possible in the first place.

We often talk about “vertical launch and horizontal scaling”. In practice, this means building one process end-to-end first, from the data source through to the application. Once that works, the same building blocks can be reused across many additional use cases.

Today, the greatest value is being created where organisations are professionalising and harmonising their data products.

Anders Landig:
And which developments are particularly exciting within the SAP ecosystem?

Mirko Pleli:
A very interesting example is the new SAP Joule Studio.

Originally, it was designed to develop custom agents for solutions such as SAP Sales Cloud and SAP Service Cloud. However, Joule Studio is evolving into a much broader platform.

Companies can now not only build agents but also automate processes, develop extensions and implement side-by-side applications. Particularly exciting is the concept of intent-based development. Here, a user describes their goal in natural language, and the platform supports the implementation.

At the same time, Joule is becoming a central access point to the entire SAP landscape. In future, users will be able to retrieve information from CRM, ERP and HR systems through a single interface without constantly switching between applications.

Anders Landig:
What does this development mean for customer experience, portals and digital customer touchpoints?

Jonas Degener:
The greatest value comes from the wide variety of interaction channels.

In the past, a portal was often the central point of access. Today, the same process can be delivered through a portal, a mobile app, a chatbot or, in the future, an AI agent.

For users, the channel itself is becoming increasingly irrelevant. What matters is being able to complete a task quickly and easily.

Mirko Pleli:
That is exactly where AI shows its strength. An assistant should not merely answer questions but support or even carry out specific tasks.

The real value comes from integration with enterprise systems and processes. That is what turns chatbots into genuine digital assistants. At the same time, portals are becoming part of a larger ecosystem rather than remaining the sole point of access to digital services.

Anders Landig:
A term that comes up repeatedly in these discussions is MCP. Why is this topic receiving so much attention at the moment?

Jonas Degener:
MCP stands for Model Context Protocol. In simple terms, it is a standard that enables AI systems to access functions and processes within enterprise systems.

This means that an agent can not only provide information but also perform actions, retrieve data and trigger processes.

The important point is that existing authorisation and security concepts remain fully intact. Through the agent, users only gain access to information and functions they are already authorised to use.

This opens up entirely new possibilities for interacting with enterprise systems.

Anders Landig:
Listening to you, one could get the impression that in the future we will work almost entirely through chats and that traditional applications will disappear. Is that the direction we are heading in?

Jonas Degener:
Not necessarily. I do not believe front ends will disappear. What will change is the way front ends are built and used.

Today, we work primarily with fixed applications. In the future, an organisation’s real capabilities will increasingly reside in services, data products and functions. Different interfaces will then be able to access the same capabilities.

The very same process could be executed through a portal, an app, a chat interface or an agent. The underlying capability remains unchanged. Front ends become more interchangeable, while business capabilities themselves move to the centre of attention.

Anders Landig:
So the workplace interface will become far more individual?

Jonas Degener:
Exactly. We are moving away from the idea of an application that looks identical for every user.

In future, an employee could receive precisely the interface they need for a particular project or use case. That interface would provide the relevant information, processes and KPIs and could adapt again as requirements change.

This creates a much higher degree of flexibility. Companies no longer need to develop a dedicated application for every use case. Instead, existing capabilities can be recombined repeatedly in new ways.

Mirko Pleli:
This is ultimately a development we already know from our personal lives. We are accustomed to content and information being tailored specifically to us.

In many business applications, that level of personalisation has only been possible to a limited extent. Users often had to adapt to the software. In future, the software will increasingly adapt to the user.

That does not mean every employee will build their own application. However, the opportunities for personalisation will become significantly greater, and with that often comes greater acceptance of the systems themselves.

Anders Landig:
So ultimately this is not about replacing people with AI but about making their work more relevant and effective?

Jonas Degener:
Exactly. From my perspective, the major change is not that people disappear. The change lies in how people collaborate with digital systems.

In future, companies will provide fewer rigid applications and instead offer data, services and functions that can be used flexibly. People will then have greater freedom to decide how they want to access those capabilities.

As a result, the digital workplace will become more personalised, more flexible and better aligned with the task at hand.

Anders Landig:
To conclude, what first step should organisations take now?

Mirko Pleli:
The first requirement is a clear target vision. Companies should define where they want to be in the long term and what role AI is expected to play.

The key is to create a foundation on which many different use cases can later be built. Organisations that only solve a single use case often end up creating an isolated prototype rather than a scalable solution.

Jonas Degener:
I would summarise it like this: build the toolbox first, then solve the use cases.

Many companies today develop a bespoke solution for every individual requirement. A more sustainable approach is to start by establishing the data architecture, integrations and reusable capabilities. Once these are in place, new use cases can be delivered far more quickly and cost-effectively.

Conclusion

The future of Business AI will not be determined solely by new language models or agent systems. The most successful organisations will be those that bring their data, processes and systems together on a common foundation. Only then can they create the conditions required for scalable AI applications.

At the same time, we are witnessing a fundamental shift in the way people interact with enterprise software. Instead of rigid applications, flexible capabilities, data products and services are moving to the forefront. Front ends are becoming more personalised, processes more channel-independent, and digital workplaces more closely aligned with the specific task being performed.

In other words, the future does not belong to a single perfect AI agent. It belongs to organisations that establish the right foundations and build a flexible digital ecosystem on top of them.