The Direct Line to the Warehouse: EPG Expands AURA with the Communicator

With the EPG AURA Communicator, EPG (Ehrhardt Partner Group) expands its AI-native environment EPG AURA with a dialogue-based access point for operational data and internal company knowledge. The new solution makes information that was previously spread across warehouse management systems, automation technology, documentation and tickets accessible via role-specific AI assistants. This is made possible by the integrated AI capability Conversation Intelligence: employees formulate their questions in natural language, while the Communicator draws on the approved sources and provides the information relevant to the specific task. Access is available via chat through the EPG Virtual Assistant EVA, or voice-based and hands-free through LYDIA Voice directly within the running process. 

When an order stalls in the warehouse or a fault occurs on an automated component, a time-consuming search across multiple applications often begins. Employees switch between system screens, search through documentation or turn to shift management — and what starts as a simple question can become an operational obstacle if the right contact person is unavailable. Particularly during deviations, shift changeovers or the onboarding of new employees, it becomes clear how heavily smooth operations still depend on detailed system knowledge and the experiential expertise of individual people. The AURA Communicator addresses precisely this gap by shifting information access to the point where the question arises. Users no longer need to know which system holds a particular piece of information, nor which screen, documentation or ticket structure to search. Instead, they formulate their query in natural language, whereupon an assistant configured for the specific task evaluates the connected sources and provides the relevant information in the context of the request. 

Individual Assistants for Different Tasks 

The technological foundation is the AURA Cognitive Core, which acts as a central intelligence and security layer, consolidating information from connected sources, processing requests and taking into account defined roles and permissions. In addition to structured data from operational systems, machine-readable documents, work instructions or system messages can also be incorporated. Companies can independently expand this knowledge base without having to set up a large-scale IT project for each new source. 

Building on this, assistants can be configured for different tasks and user roles. Companies determine which information each assistant may access, who can use it and in what format responses are delivered. This means warehouse employees receive different information than, for example, shift management, service staff or administration. In addition to EVA and LYDIA Voice, Microsoft Copilot can also be connected as an access channel, making the assistants available within an existing Microsoft environment. 

Making Operational Systems Directly Accessible 

Conversation Intelligence goes beyond searching documentation and stored company knowledge. Via standardized interfaces, the AURA Communicator also unlocks information from operational systems, enabling queries in natural language for data such as inventory levels, storage locations, order statuses or current process states from warehouse management systems. In addition, warehouse automation systems, shuttle solutions, automated guided vehicles and individual automated components can be connected, as can applications for documentation, knowledge management or ticketing. Existing third-party systems can also be integrated without replacing the current system landscape, with each application continuing to serve as the leading system for operational processes. Users can thereby not only determine where a specific item is stored, but also understand why an order is not being processed further. In the event of a technical fault, it is simultaneously possible to check whether a ticket or a suitable work instruction already exists. Information that previously had to be searched across multiple applications is thus immediately available within the relevant process context. 

Secure Information Access in the Enterprise Environment 

Role-based permissions ensure that users receive only the content approved for their tasks, while companies can define which sources and systems an assistant is permitted to access. The required security architecture is provided by the Cognitive Core, which protects the application against manipulation attempts such as prompt injection, among other threats. Direct information access also relieves the burden on operational staff and shift management, as recurring questions about inventory levels, process states or work instructions no longer need to be resolved through individual subject matter experts. Natural language interaction also simplifies access to complex system landscapes and, with support for more than 60 languages, reduces language barriers in international teams. 

Recognizing Connections Across System Boundaries 

In a next development stage, the Cognitive Core is set to correlate structured operational data even more comprehensively with unstructured information. Using graph databases, current system states could then be matched against existing tickets, technical documentation and previous fault cases. The AURA Communicator is therefore intended to not only surface individual status information in the future, but to make connections between different sources visible and derive possible recommendations for action. Information will no longer need to be sought where it is stored, but will instead be available at the moment it is needed for a decision or the next process step.