Kick-Off Digital Filing Cabinet
Katharina Kloiber | 12.06.2025
Together with Christoph Döberl (zynd GmbH), we are evaluating how AI systems and models operate on local hardware in a high-performance, reliable and cost-efficient manner. AI systems are on everyone's lips, but their use in companies must always be handled with care, especially when internal company information is involved. One of the most important factors when introducing AI in companies is data security. Many well-functioning AI systems are based on either transferring data to central systems or consist of locally installed models that constantly communicate with central systems. This is a security risk that NO company can take.
On the other hand, companies are under increasing pressure from their employees and the market, as AI often leads to increased efficiency and makes tasks easier, e.g. programming, writing texts, preparing quotations or conducting research. However, if this takes place outside the company structures and secure infrastructure, it results in a loss of information and control for the companies.
Host AI models on your local hardware
We are working to counteract this situation. Installing local AI systems that have no connection to central systems of thirds can be realised with the right hardware. Keeping data and models completely and exclusively local is therefore affordable and has added value: implementation on local, self-controlled hardware preserves companies' data sovereignty.
In order to use modern models and agents, targeted hardware is required. We evaluate the extent to which a high-performance and sustainable in-house solution can be established by cascading affordable systems. Security plays an important role here, which is why the system is built purely on open source operating systems and components.
