Dream team: AI and ERP
Stefan Hillman is a software architect at KUMAVISION. In an interview, he explains why AI perfectly complements the functionality of the ERP system and why artificial intelligence is also an issue for medium-sized companies.
How can artificial intelligence generally support an ERP system?
Stefan Hillman: All processes and data of a company typically come together in the ERP software. Artificial intelligence makes it possible to check these amounts of data for patterns and structures and to correlate them with each other. The use of AI is always useful when it comes to increasing efficiency through automation, increasing process reliability, creating more transparency through better insights and reliable forecasts, and developing and establishing new business models. In addition to machine learning, cognitive AI services such as speech, image and text recognition, process automation and virtual assistants are available for company-wide use.
How can AI functionalities in ERP contribute, e.g. B. to optimize the storage processes?
Stefan Hillman: A first approach is to use machine learning to optimize the selection of storage locations depending on picking behavior and to take cross-selling relationships recognized and learned by the system into account. Items that are often bought together are arranged in such a way that they can be picked one after the other without having to walk long distances. Optimized routes save time and thus contribute to reducing costs. The use of AI-supported image recognition increases process reliability, since confusion between similar-looking items is detected at an early stage. It can also be used to monitor compliance with packing regulations. Some companies work with Pick-by-Voice. The warehouse worker receives the picking instructions via a headset and has to confirm them with fixed commands. AI makes it possible to dispense with these fixed commands and instead work with natural language. But AI also opens up numerous possibilities in procurement: For example, with automated suggestions for minimum and maximum stock, ordering strategies that take delivery times and price developments into account. If, in addition to the ERP system, external data sources such as e.g. B. If suppliers' factory holidays are taken into account, the data quality increases.
When and how did you as an ERP provider discover the disruptive potential of AI applications and start integrating AI functionalities into your software?
Stefan Hillman: We not only look at AI from the point of view of disruption, but also from the point of view of continuity. Microsoft Dynamics has had AI functionalities since 2016/2017. The roadmap of our ERP industry solutions envisages numerous other AI-based functions for 2021, which we are currently developing. On the one hand, this involves further developing existing intelligent algorithms through machine learning. On the other hand, we are also laying the foundation for disruptive business models and services. Topics such as AI clearly show that the role of the ERP partner is changing fundamentally: Providers are increasingly becoming an innovation companion who has to cover a wide range of technology, process and industry expertise. The aim is no longer just to offer software, but to enable customers to develop and offer new services. In addition to know-how in areas such as AI or IoT, agile project methods are also required.
What were the special technological requirements / obstacles that had to be overcome (e.g. requirements for computing power, data volumes, etc.)?
Stefan Hillman: Our technology partner Microsoft provides a range of AI services that integrate seamlessly into our ERP industry solutions on the one hand and act as a Cloud-Let the offer scale easily. We are therefore in the fortunate position of being able to fully focus on the development and implementation of customer and industry-specific AI solutions without having to worry too much about technological obstacles. The challenge is rather to define the requirements as precisely as possible together with the customer. Because the answers that an AI-based solution provides can only ever be as good as the underlying question. Meaningful AI models require data. What sounds banal has a serious background. In the past, not all required data was always collected due to performance and storage space limitations. Apparently irrelevant data in particular often has great potential for establishing connections and gaining deeper insights. In AI projects, it is therefore often necessary to first collect data with which the system can then be trained.
Medium-sized companies and AI - can they go together? What do you say to customers who think they are “too small” for that?
Stefan Hillman: AI is not a question of company size, but a question of attitude. It always depends on the specific question that is to be solved with AI. For example, while the automatic translation of receipts can be easily retrofitted as an app in Microsoft Dynamics, automated damage detection for rented machines requires more effort. Companies don't have to start from scratch with AI. At KUMAVISION we have many years of experience in various industries, and we are familiar with the industry-specific issues and the corresponding basic AI models. These best-practice solutions often serve as a starting point, which is then adapted to customer-specific requirements and trained with your own data.
The integration of AI elements requires a change in technology - in your experience do users react skeptically or openly? What are the prejudices (if any) that users may have towards AI?
Stefan Hillman: AI provides answers to questions that cannot be answered manually or only with great effort. In this respect, AI is initially perceived positively. Occasionally, privacy and data security concerns are raised. Today, AI models can be trained online and then used offline. This means that in live operation, processing takes place locally, no data is transferred to third parties. Employees sometimes perceive AI as a threat to their jobs. It is exactly the opposite: the knowledge and experience of the employees are indispensable to create an AI model and to classify the results. AI relieves employees of time-consuming routine tasks and thus enables them to concentrate on the actual core tasks.
What are the next big steps in the ERP area with regard to the integration of AI applications?
Stefan Hillman: At KUMAVISION we see two topics with RPA and virtual assistants that will change the ERP landscape. Processes are becoming increasingly complex today and require numerous documentation requirements. Closer networking with an increasing number of business partners, suppliers and service providers also requires compliance with a wide range of specifications, which means that employees spend more and more time maintaining processes. At the same time, however, costs are to be saved. Robotic Process Automation (RPA) makes it possible to automate processes and make decisions independently. Virtual assistants promise further relief. An example: When going on a business trip today, you have to enter an appointment in Outlook, create an order in the ERP, book a hotel and reserve a company car. In the future, a virtual assistant will ask for the key data in natural language in a dialog and independently initiate the necessary processes in the various applications. The employee therefore has a single point of contact, the assistant takes care of everything else in the background. Artificial intelligence will radically change and simplify the user interface of ERP systems in the coming years. Voice control, dialogs with virtual assistants, automation in the background and many other AI services will make operation easier in the long term.
What opportunities does AI offer your company?
Stefan Hillman: Together with you, our experts identify opportunities and possible uses of artificial intelligence in your company and show you which AI applications will bring you further. Make an appointment now for a personal consultation: