Zenta Solutions Tests Medical Data Structuring AI on the Leonardo Supercomputer via AI Factory Austria

AI Factory Austria (AI:AT), Austria's national platform for applied AI and high-performance computing, announced in November 2025 its first practical case: helping the Viennese company Zenta Solutions gain access to European supercomputing resources. AI:AT is implemented by the AIT Austrian Institute of Technology together with Advanced Computing Austria (ACA), is part of the European AI Factories initiative supported by the EuroHPC Joint Undertaking and Horizon Europe, and acts as a one-stop shop for technical consulting, computing resources and training. Zenta Solutions, founded in 2025 as a spin-off of the Medical University of Vienna, develops AI for structuring and analysing medical data. With AI:AT's help it became the first company to use the Leonardo EuroHPC system in Bologna, Italy, where its AI-based processes are being tested and optimised. "The first step is to completely anonymise the data. It is then analysed, contextualised and converted into a structured form using image-based AI models and large language models," said David Haberl of Zenta, adding that this turns unstructured health data into a basis for applications such as predicting treatment response. Zenta relies on open source models that can run entirely on the user's own systems without external cloud infrastructure, so sensitive health data stays on site. The work on Leonardo is still at the testing stage: "We are currently testing our methods on Leonardo using open data sets and realistically simulated sample data," Haberl said. Co-founder Thomas Beyer said the company got access to the EuroHPC systems within a few days and can develop its models without building its own expensive infrastructure. AI:AT says it is building up its service portfolio step by step, with a focus on helping small and medium-sized companies reach high-performance computing, and is already in contact with other companies with similar needs. AI:AT is co-funded by the EuroHPC Joint Undertaking under grant agreement no. 101253078.