T-Systems presents on t-systems.com the AI chat, a Retrieval-Augmented Generation (RAG) assistant that provides users with precise, contextual, and up-to-date information, and improves the website experience. The AI chat combines proprietary solutions and platforms and was developed with a focus on sovereignty, security, and user-friendliness.
The AI chat is a standalone solution by T-Systems. It is based on internal components such as AI Foundation Services, the LLM Hub, T Cloud, and Telekom’s Open Design System (ODS). This enables independent development without relying on third-party providers. The AI chat not only reduces search time, but also enables targeted lead generation and relieves the burden on T-Systems’ customer service, as questions are answered directly and relevant content is highlighted in a targeted manner. T-Systems thus demonstrates not only AI expertise, but also the ability to implement digital solutions in a practical and secure manner.
The answers draw on a dedicated knowledge base of editorial texts, uploads, and crawled web pages. Sources are cited transparently, and regular re-crawling keeps the data up to date. The AI chat is built on T-Systems terminology, ensuring that communication remains consistently professional. This prevents misunderstandings and provides clear, industry-relevant information.
In the beta phase, the AI chat supports German and English. In the future, the AI chat is also to be integrated into the websites of local business units for international customers, and thus expanded to include additional languages such as French, Spanish, or Dutch.
Security is at the forefront: The AI chat runs on the certified public cloud solution from T Cloud and is protected by 2‑factor authentication. Intelligent guardrails additionally prevent misuse. The model is not trained with end-user chat logs; GDPR requirements are implemented both technically and organizationally, for example through pseudonymous IDs in the public chat and encrypted data transmissions. Internal product security analyses ensure compliance, while deletion pathways and detection patterns intercept sensitive data.
With a modular architecture, the AI chat continuously adapts to new requirements. Depending on the use case, the system draws on various large language model (LLM) instances. Performance is measured by API response times, error rates, and user satisfaction. Dashboards with trend analyses ensure transparency. User feedback flows directly into improvements, while automated processes ensure continuous updates.
The AI chat promotes an intuitive user experience through suggested follow-up questions, direct rating buttons, and a consistent ODS design. The interface is clearly structured, supports switching between German and English, and enables quick orientation. Users can provide feedback or report errors or ambiguities at any time—one click is all it takes. This supports further development as well as quality assurance.
With the AI chat, T-Systems is betting on innovation, security, and user orientation. The digital assistant is an example of the combination of group expertise and modern technology—a step that strengthens both website performance and customer trust.
