A new era is currently dawning for user interfaces. The introduction of artificial intelligence (AI) into our everyday lives has led to increased expectations of business software: employees now expect their business applications to offer the same intuitive, dialog-oriented interfaces that they also use privately.

This has led to a comeback of character-oriented user interfaces and a return to interaction via text input and output.

Many applications use pure text and can be used intuitively and without instructions for a specific goal. However, texts can only convey the structured data that is common in companies to a limited extent. And without real-time updating, static, text-based results become less relevant as soon as they are generated.

Structured data is easier to process when users can filter, sort and visualize it. For this reason, graphical user interfaces (GUIs) are ideal for displaying structured data and guiding users through complex workflows. But developing GUIs is expensive, leaves little freedom, and requires generic, one-size-fits-all solutions that make it difficult to deliver the dynamic, tailored user experience that users demand today.

Text is flexible but has limited possibilities; GUIs, on the other hand, are powerful but inflexible. “Generative UI” – the concept of AI-powered user interfaces – bridges these two approaches and opens up new opportunities in the field of enterprise software.

From static dashboards to dynamic workspaces

As an example, consider a scenario in which a procurement manager is investigating a disruption in the supply chain. Instead of manually reconciling data from five different applications, she can work with a prompt: “Show me which suppliers in Southeast Asia are at risk and create alternative sourcing scenarios.”

Based on this request, agents start working in the background. They collect and analyze live data, simulate results and calculate the predicted impact of each alternative. Executive agents are also available to act on orders.

Users do not have to deal with these complex processes. A dynamic user interface is available to them within seconds – not a generic dashboard, but a control center created specifically for them. Interactive maps highlight affected regions and supply chain diagrams update in real time. When users change the parameters, the risk assessment is immediately adjusted. Additionally, built-in controls are available to trigger orders or notify suppliers. This enables users to make decisions and take actions. It also simplifies collaboration: Colleagues can participate in an interactive workspace without the need for briefings or meetings to explain the context.

And this is what the future looks like: A business suite in which user intent shapes its user interface and their decisions determine the corresponding actions. To make this possible, we combine our AI solution Joule and Joule Agents with our vision for “generative UI”. It's not just about demand-oriented dashboards, but about controlling business processes using interfaces that adapt to the user's role, context and tasks. This is how “Vibe Coding” – a programming method that uses natural language prompts – can be used for business processes. The focus is shifted from syntax to intent.

We are moving into an era where user interfaces are dynamically created by AI, allowing users to instantly interact with those interfaces. Generative interfaces enable the transition from static software suites to highly customized applications that are built like temporary control centers and tailored to a specific problem.

How SAP overcomes challenges

In order to deliver an intent-driven, enterprise-wide business suite, complex requirements must be implemented. We develop generative surfaces because we understand both their potential and the associated risks – and we have the necessary expertise to address this challenge.

accuracy

Large language models (LLMs) can produce plausible but incorrect results or “hallucinate”. A chatbot that hallucinates the plot of a movie isn't a big problem; However, a procurement system that misrepresents supplier conditions has real consequences. Our approach to “generative UI” aims to directly visualize data from capture systems with transparent information about their origin. Reliable real-time data as the basis of user interfaces is the first step in avoiding inaccuracies.

Trust

How can users be confident that the user interface is reliable when every interface is dynamically generated? Trust is based on consistency and predictability. Our generative interfaces are based on the well-known and proven architectural principles of SAP Fiori for lists, dashboards and workflows. The content is tailored and the structure is consistent and familiar, allowing users to assess and respond with confidence.

complexity

Enterprise software includes sophisticated and individual functions. It has been developed over decades and is based on extensive specialist knowledge and extensive business logic. Generative UI builds on Joule's proven integration and coordination capabilities, which already connect systems within an IT landscape and control agents to execute complex workflows. Based on this, users can carry out tightly integrated processes via simple user interfaces, while Joule takes care of the necessary coordination.

Why this development is important

The private use of AI has led to increased user expectations. And the gap between the AI ​​that employees use in everyday life and the AI ​​offerings in their workplace is increasing.

The future of enterprise software is not about inflexible chatbots on old-fashioned user interfaces. Instead, the trend is toward tailored control centers – interfaces that adapt to a user's intent and are based on live data, executed by agents and controlled by users.

In this way, we can completely redesign workflows.

Jonathan von Rüden is Chief AI Officer at SAP SE.