T-Systems' AI agent platform "Talk to your data", which makes corporate data from different sources accessible via chat, reduced the time for complex analyses from six hours to less than 60 minutes in initial pilot projects, according to a March 6, 2026 Deutsche Telekom press release.
Information in companies is spread across dozens of databases and applications in different formats. The release cites studies by McKinsey and the Microsoft Work Trend Index showing that employees spend around a quarter of their working day simply searching for information. "Organizations do not suffer from a lack of data, but from the fact that their information is stored in separate systems," says Sven Giesselbach, AI expert at T-Systems.
Employees ask questions in natural language. The AI agent analyzes the request, automatically searches the relevant systems, and delivers a structured and traceable answer within seconds, optionally including charts or visualizations. At the core is an ontology layer, a semantic layer that connects different data sources and creates shared meanings between pieces of information even when they are labeled differently across systems. "Together with our customers, we build this ontology and link the relevant data sources," explains Giesselbach.
The release describes real-world testing in hospital environments, where diagnoses, lab results, medication plans, and medical reports are often stored in different systems. With "Talk to your data", doctors can retrieve this information directly: a single query is enough to access all diagnoses of a patient or to display the development of vital signs over several months in a chart. The agent automatically sends multiple queries to different systems, evaluates the results, and checks whether additional sources might be relevant. It can also point out possible inconsistencies, for instance when a laboratory value does not match the documented therapy.
"Talk to your data" runs on T-Systems' Agentic AI platform and uses its AI Foundation Services, which provide models, infrastructure, and security mechanisms. T-Systems operates it entirely within its own cloud infrastructure, so sensitive information never leaves the protected environment. The Industrial AI Cloud provides the required scalability and supports both large language models and more specialized, efficient models.
"What previously took around six hours can now be completed in less than one hour with the intelligent assistant," Giesselbach reports, describing a public-sector project in which extensive reports had to be researched and prepared.