Press Release
Lindau, September 24, 2026
With Semantic AI, PANTOPIX SPHERE combines knowledge graphs, large language models, and flexible data integration into a unified knowledge base for enterprise AI.
Retrieval-Augmented Generation (RAG) has established itself as an important approach for connecting large language models with a company’s proprietary information. However, when dealing with complex product and company knowledge, simply finding relevant documents or text passages is often not enough. The relationships between pieces of information are also crucial: Which component belongs to which product variant? Which documentation applies to which market? And how are products, features, processes, and technical information related to one another?
GraphRAG extends classical RAG by adding this level of relationships. In addition to relevant content, the approach also takes into account semantically linked knowledge from a knowledge graph. A benchmark for enterprise data demonstrates the impact this additional context can have: GPT-4’s accuracy increased by 40 percent when queries were run via a knowledge graph representation rather than directly on the underlying SQL database. With the PANTOPIX SPHERE Knowledge Platform, PANTOPIX therefore connects large language models with explicitly modeled entities, facts, and relationships. This ensures that AI applications receive not only relevant information but also the domain-specific and company-specific context.
A large language model is very good at handling language and recognizing similarities. However, it does not automatically know which relationships actually apply within a company from a business perspective. To create robust enterprise AI, we must therefore combine the flexibility of LLMs with a reliable knowledge structure.
Semantic AI as the Foundation for Connected Corporate Knowledge
PANTOPIX SPHERE is based on the principle of semantic AI: large language models process language and unstructured content, while knowledge graphs explicitly and transparently map relationships. Ontologies and taxonomies define the company-specific knowledge model, such as business objects, technical terms, and their relationships.
At the same time, PANTOPIX SPHERE integrates information from various sources. Structured data from databases and line-of-business systems can be integrated, as can documents, metadata, or taxonomies. Through APIs, the Model Context Protocol (MCP), and preconfigured connectors, the knowledge base can be connected to existing systems and AI applications. Role-based permissions and data lineage ensure that access to information, as well as its origin and processing, remain traceable.
The technological advancement of PANTOPIX SPHERE is also the subject of a funded research and development project. In July 2026, PANTOPIX received a federal research grant for this purpose. Among other things, the project focuses on the interaction of large language models with rule-based approaches, as well as the prospective control of information flows via natural language.
For us, GraphRAG is not an AI feature that was added as an afterthought. PANTOPIX SPHERE was designed from the outset to semantically link information across system boundaries. Generative AI can be deployed on this knowledge base exactly where its strengths lie.
Interview Request
What are the limitations of traditional RAG in a corporate setting? When does GraphRAG offer added value? What role do knowledge graphs and ontologies play in ensuring reliable AI applications? And how much corporate knowledge can actually be extracted automatically?
Maximilian Gärber is available to speak with editorial teams for interviews and in-depth discussions on GraphRAG, semantic AI, knowledge graphs, and, of course, PANTOPIX SPHERE.
Press photos:
About PANTOPIX
Through semantic models, knowledge graphs, and AI-based approaches, we create intelligent information systems that maximize the value of digital product data. Our portfolio includes product development, consulting, and implementation of modern systems—always in close collaboration with our customers to deliver sustainable, customized solutions.
