AI Applications in SAP EWM
Introduction
The integration of artificial intelligence into warehouse management is no longer just a vision of the future. In recent years, SAP has made significant investments in integrating AI functions into its product range – and SAP Extended Warehouse Management (EWM) is no exception.
In this article, we highlight the current possibilities for using AI in SAP EWM – from natively integrated AI functions to integration scenarios via the SAP Business Technology Platform (SAP BTP). We also present ideas for possible application scenarios for AI in SAP EWM that can realistically be implemented today.
Native AI Functionality in SAP EWM
In the latest releases of SAP S/4HANA EWM, various AI and machine learning functions have been integrated directly into EWM. These functions can be used without additional third-party solutions and integrate seamlessly into existing processes.
Slotting by Machine Learning
SAP S/4HANA 2022 Feature Pack 1 introduced the Slotting by Machine Learning function. This function extends slotting and uses machine learning algorithms to automatically suggest optimal warehouse product master data for new or changed products.
It works as follows: SAP EWM analyzes the master data settings of existing products and their warehouse tasks, e.g., for picking, and derives patterns from them. Based on these patterns, the system automatically suggests the following warehouse product master data:
- Storage control indicator
- Storage area indicator
- Storage bin type indicator
The advantage: You significantly reduce the manual effort involved in master data maintenance and benefit from more consistent storage strategies, as the system automatically learns from proven patterns. However, a prerequisite for meaningful application is that you already have warehouse products that are correctly or optimally maintained.
Predictive Labor Demand Planning by Machine Learning
Predictive Labor Demand Planning has been available since S/4HANA 2023 Feature Pack 3. The solution is based on machine learning algorithms and is only available for S/4HANA, Private Edition, i.e., only in the cloud. The functionality analyzes historical workload data records and uses them to calculate the expected duration of future warehouse orders and activities. This enables department and shift managers to estimate staffing requirements earlier and more accurately.
ABC Analysis and Data-based Classification
SAP EWM offers another function for data-based optimization with its integrated ABC analysis. ABC analysis automatically categorizes products based on historical stock movement data. The classification of the warehouse product determined in this way can be used in combination with slotting to automatically move fast-moving items (A items) to ergonomically favorable storage types and storage areas, for example.
Joule – SAP's AI Co-Pilot
With Joule, SAP has introduced an AI assistant that is now integrated into numerous SAP applications in the private cloud and public cloud. Joule enables interaction with SAP systems in natural language and provides support for navigation, data search and evaluation, and task execution.
Several specialized Joule agents have been announced for supply chain applications, including a Production Planning and Operations Agent that can check material availability and capacities and validate production orders. Although these agents are primarily aimed at production planning, they show the direction in which AI is moving in logistics.
AI Integration via SAP Business Technology Platform
In addition to the natively integrated AI functions, the SAP Business Technology Platform (SAP BTP) provides a powerful platform for integrating external AI services and APIs. Using SAP AI Core and the Generative AI Hub, you can integrate both SAP's own and external large language models (LLMs) into your business processes.
ABAP AI SDK and Intelligent Scenario Lifecycle Management
The ABAP AI SDK and Intelligent Scenario Lifecycle Management (ISLM) provide a framework that enables the integration of generative AI directly into ABAP applications.
The most important features of the SDK for developers:
- Integration of large language models (LLMs) into existing ABAP applications
- Creation of prompt templates for recurring AI queries
- Connection to the SAP Generative AI Hub
- Lifecycle management for AI scenarios with transport and versioning
SAP Generative AI Hub and Cloud Connector
The Generative AI Hub in SAP AI Core provides access to various LLMs. These models can be accessed from SAP applications via API, making this the preferred integration method for using external AI APIs in your SAP S/4HANA EWM on-premise or private cloud. For a secure connection to SAP BTP, you also need the SAP Cloud Connector. This allows existing EWM systems to benefit from AI services in the cloud without compromising the security architecture.
Specific Use Cases for AI in SAP EWM
The technical possibilities are endless – but many companies are looking for specific use cases for AI in warehouse management. Below, we present four examples of use cases for artificial intelligence in SAP EWM that can already be implemented today and illustrate the potential for using AI in SAP EWM.
The Challenge
Traditional putaway strategies primarily take physical product characteristics such as size or weight into account. In SAP EWM, the putaway control indicator in the warehouse product is usually maintained manually and statically to define the putaway strategy. This does not take into account which products are frequently ordered and delivered together. This leads to longer picking routes and inefficient processes – in the worst case, either additional consolidation or separate shipping is required later in the process. Both are significant cost factors and should therefore be avoided if possible.
The Solution
Machine learning algorithms can be used to identify patterns in historical order data. The system learns which products are frequently ordered together and uses this information to influence the storage strategy, e.g., by adjusting storage control indicators or storage area indicators. Products with a high correlation are thus stored close to each other.
Implementation in SAP EWM
The connection to the machine learning engine is made either locally on a separate server or via SAP BTP. In these cases, an external ML engine is used. In theory, the machine learning algorithms can also be implemented directly in SAP EWM.
The storage strategy can be adapted in many ways, depending on the specific problem and existing customizing of the storage strategy in SAP EWM. In addition to the direct, automatic maintenance of storage control indicators and storage area indicators, the implementation of a correlation ranking is also conceivable, so that an extension of the storage strategy can directly access correlated products during storage location determination. This extension works similarly to the putaway rule “Putaway near fixed bin,” whereby the equivalent of the fixed bin in this case is the storage bin of a correlated product.
The Challenge
In goods receipt, deliveries often have to be created or adjusted manually based on physical, paper-based delivery notes. This is especially true if you have set up the goods receipt process based on expected goods receipts (EGR). However, even for existing deliveries, the printed delivery note data must be reconciled with the data in the system.
This is time-consuming and prone to errors.
The Solution
An AI model for image recognition can extract the relevant data from photos or scans of delivery notes and automatically compare it with the expected goods receipts or existing delivery documents. If there is a match, the delivery is automatically created or, if necessary, the delivery quantity is adjusted in accordance with the tolerance settings in the system.
Implementation in SAP EWM
The technical starting point for this process is either the use of a mobile dialog based on SAPUI5 or an extension in the desktop transaction /SCWM/GRPE. A photo or scan of the delivery note is taken and sent to the AI API via Cloud Connector through the Generative AI Hub in SAP BTP. The AI for image recognition can also be operated and connected locally if there are data protection concerns.
The AI determines the relevant data for delivery creation from the image data and returns it to SAP EWM. Alternatively, the AI can also determine the data itself, search for a suitable EGR or delivery in EWM via API, and trigger delivery creation or quantity adjustment via API. The search for suitable EGRs and inbound deliveries can be carried out using the order number specified on the delivery note or (if inbound deliveries already exist) the delivery notification number or, if necessary, the vendor number and material number.
The Challenge
Quality inspections in goods receiving, in the warehouse, or during returns processing are often time-consuming and heavily dependent on the experience of individual employees. Damage, incorrect labeling, or deviations from the target condition are not always reliably detected.
The Solution
With the help of computer vision and image recognition algorithms, incoming goods, returns, or items in the warehouse can be automatically checked for quality criteria. The technical implementation is carried out either by cameras at strategic points in the warehouse or via a handheld device (MDE). The images are then analyzed by AI in real time and the quality assessment is returned.
Implementation in SAP EWM
Integration can be achieved in various ways. For example, an AI service for image recognition can be connected via API using Cloud Connector in SAP BTP. The results of the image analysis are then returned to SAP EWM or forwarded by SAP BTP to another AI service that calls the SAP EWM APIs to, for example, make a usage decision for the inspection lot or inspection document or for the return, set a status on the handling unit, or transfer the stock to blocked stock.
Possible inspection criteria could include damaged packaging, the legibility of labels and barcodes, or other anomalies (e.g., discoloration).
The Challenge
The warehouse management monitor in SAP EWM displays a wide range of data: open warehouse tasks, storage bin capacities, handling units, delivery and shipment items, and message queues. However, interpreting this data and deriving recommendations for action can require a great deal of experience and time.
The Solution
By adding an AI interface to the warehouse management monitor—for example, as a generic monitoring method—employees can work with AI to interpret the displayed data in natural language. An LLM analyzes the data in context and provides recommendations for action.
Implementation in SAP EWM
Integration in the warehouse management monitor is achieved, for example, via a monitor method that performs an API call to an AI service. The API call can be made via Cloud Connector using the Generative AI Hub in SAP BTP or via direct API integration. The monitor data and the user's question are transferred as context, and the AI response is displayed in the monitor.
This allows users to utilize AI as a decision-making aid when prioritizing warehouse tasks or identifying bottlenecks, for example.
Prerequisites for AI Use in SAP EWM
Before you can use AI functions in your SAP EWM system and implement the scenarios described here, you should check the following prerequisites:
- System requirements: For the native ML function Slotting by ML, you need at least SAP S/4HANA 2022 FPS01.
- Licensing: Slotting is part of SAP EWM Advanced and therefore requires a separate SAP EWM license.
- Connection of SAP EWM to BTP: To use the ABAP AI SDK and the Generative AI Hub, you need an SAP BTP instance with the appropriate service agreements.
- Data quality: Machine learning algorithms and generative AI are only as good as the data they are trained on. A clean master data base is crucial for the success of your AI project.
- Protecting company data: Before implementing any of the AI use cases for SAP EWM described here, you should check whether your data should actually be sent unmasked to the respective data processing center, i.e., to the AI API. You should consider data protection directly in your implementation and, if necessary, fall back on locally operated AI instances.
- Clean Core first: Where possible and appropriate, use the Clean Core principles for developing your extensions and control your AI integration for SAP EWM via SAP BTP.
Conclusion
AI in SAP EWM is no longer a topic for the future: the technological foundations are already in place today and offer a wide range of possibilities for integrating AI into your warehouse processes.
However, the key to success does not lie in the technology alone; you should first clarify which use case makes sense for your company and your warehouse logistics. Only AI integration that achieves actual quality improvements and/or cost savings is really worth the effort in the end. We hope that this article has given you a little inspiration to get creative yourself.
Your Next Move
Would you like to find out which AI use cases are suitable for your warehouse? Our SAP EWM consulting services will help you evaluate the potential of AI in your SAP EWM and identify further use cases.
