A Pipe Dream or Reality? Optimizing Project Management in SAP with AI

Everyone is talking about artificial intelligence; research and development are proceeding tirelessly, and AI is rapidly finding its way into everyday business life. As one of the first SAP consulting firms, we plan to Milliarum, to make AI-based applications available as early as this year, including in the area of SAP project management. In this interview, our CEO, Dirk Ott, offers insights into the potential of artificial intelligence in the development of SAP add-ons and consulting services.

How and in what areas can artificial intelligence already be used today in SAP add-ons and in consulting?

Dirk Ott: The use of artificial intelligence in SAP consulting is nothing new in and of itself. For example, we’re all familiar with SAP CoPilot, which was already able to access SAP functions via voice commands. That service has since been discontinued and will be replaced by new cloud-based chatbots from SAP that are currently under development. In contrast, at Milliarum, we focus on providing chatbots for on-premises applications. We work with sensitive, company-specific data. The advantage of on-premise solutions—particularly with regard to data protection concerns—is that they are in-house solutions installed on the company’s own servers and can be managed from there. This gives our users full control over their data.

And what might the use of AI look like in the future?

Dirk Ott: Our vision is for every SAP expert to be able to specialize in their core area. However, this expertise will not be limited to areas such as controlling or SAP project management. With the help of artificial intelligence, it will also be possible to quickly acquire detailed know-how in other areas of SAP applications or with entirely different tools. The trend will be for experts in their respective fields to be able to define even complex system requirements. Artificial intelligence is revolutionizing the use of SAP, which is generally considered complex: it will be significantly simplified. Everyone will be able to program and operate complex systems. This will significantly reduce the need for training. The only requirement is knowing the relevant terminology so that users can formulate targeted prompts and inputs in natural language. Ultimately, this will greatly increase productivity and satisfaction.

How does Milliarum use AI, and what are its vision and next milestones?

Dirk Ott: Our focus is on offering AI-based solutions to increase our users’ productivity. We are currently conducting extensive research in this area and have drawn up a rough roadmap. We’re focusing on the SAP areas where we have deep expertise—namely, SAP PS, SAP PPM, resource management, and the development of modern Fiori apps with UI5 interfaces. In doing so, we’ve identified where AI offers our users opportunities for optimization. This primarily involves simplifying the use of the SAP system. In these core areas, we will provide SAP-based chatbots that can perform specific functions using natural language.

How would this work in practice?

Dirk Ott: Take the PPM chatbot, for example—it will make it possible to create complex portfolio elements for project ideas or project proposals in a completely intuitive way. These proposals are typically created in a web interface with numerous fields. With our chatbot, it will soon be possible to use voice commands to tell the system: “Please create a portfolio element for a project idea with a specific description and defined deadlines.” The exciting part is that the AI will then be able to ask targeted follow-up questions. Anyone familiar with SAP knows that a complex object cannot be created with just a few superficial pieces of information. The AI will then actually be able to ask what’s missing and bring the whole process to completion.

What are the benefits of using artificial intelligence in SAP applications?

Dirk Ott: AI makes work significantly easier, especially for people who don’t work in SAP on a regular basis. You can also ask questions to improve usability: “What functions do I actually have at my disposal?” The AI then lists the corresponding transactions and services. Or: “Show me the ten projects or portfolio items with the highest actual costs or the highest contribution margin.” There will also be AI applications in the SAP Project System, such as cost and schedule analyses, as well as editing functions. We’re also planning a chatbot for our resource management solutions. For example, department heads could use it to identify the employees with the highest workload or to determine how their department’s capacity will be utilized over the next twelve months. In doing so, the AI will not only display textual information but also generate clear graphics. When developing Fiori apps, AI-powered tools can assist developers by generating ABAP code, thereby accelerating the development process.

Is this all just a pipe dream? When can users actually expect to start using these tools?

Dirk Ott: The first requirement is that the technical infrastructure be in place in SAP. We plan to launch the first SAP chatbot in on-premises solutions as early as the second quarter of 2024. In addition, our roadmap calls for the PPM and PS chatbots to be implemented in the third quarter. They will be connected to our solutions in the background via an OpenAI interface. This will make it possible to operate the tools via ChatGPT or an API interface. However, this is still primarily cloud-based. That’s why we’re simultaneously working on local models that can be used on the corporate network. In the future, we’ll also be able to integrate proprietary AI models into our infrastructure. In doing so, we’ll fundamentally support all solutions that gain traction in the market. Resource management will follow in the fourth quarter. The goal is to lower the barrier to entry for SAP and enable a faster and more productive way of using the system.

What are the concerns, challenges, and limitations?

Dirk Ott: The concerns primarily relate to data protection, as already mentioned. Another current problem is what are known as “hallucinations”—that is, statements made up entirely by the chatbots. These can lead to serious consequences and must be avoided. We address this by ensuring that our chatbots access only data from the company’s internal SAP system via API interfaces. Another challenge is the hardware: High-performance artificial intelligence requires massive server farms, which are currently popping up everywhere like mushrooms. Local models require very powerful hardware. Training is still extremely time-consuming, resource-intensive, and expensive. Milliarum ensures that hardware requirements are kept as low as possible when integrating local models. Meanwhile, a great deal of research is still being conducted to reduce the AI’s resource consumption, and we look forward to seeing how things develop.

Thank you very much for the fascinating conversation!