ITZBund and SAP Bring AI to German Federal Agencies

A universal artificial intelligence (AI) platform for all federal authorities: that’s what the German Federal Center for Information Technology (ITZBund) wants to create – relying, among other things, on a scalable modular system based on SAP HANA and SAP Data Intelligence. The AI platform has already gone live at one of the federal agencies.

It all began in 2015 with a cabinet decision by the German Federal Office of Administration. The IT landscape was to be consolidated and homogenized by a single IT service provider, the ITZBund.

The mission: shape the future digital transformation of the German federal government and its administration. The ITZBund operates 37,200 servers in its own data center and manages more than 100,000 end devices, consolidating disparate IT solutions across all German federal agencies.

Artificial Intelligence for the German Federal Office of Administration

The German government’s coalition agreement highlights AI’s importance in the coming years and names it one of the most disruptive technologies. The ITZBund worked with SAP to create an AI platform on which the first AI projects can be implemented. Based on SAP HANA and SAP Data Intelligence, a modular and scalable AI platform was created, bringing together data provision, model development, and model application in a single solution. “For the very first time we have now systematically introduced AI to the German Federal Office of Administration,” says Holger Lehmann, management team leader in the ITZBund.

AI-supported methods can, for example, predict economic developments, analyze large volumes of data, and detect anomalies. AI capabilities can also uncover links and patterns in images and texts. There are numerous use cases in which AI aids specific federal agencies. Examples include automatic fraud detection, monitoring and controlling traffic jams, detecting storm damage in forests, and generating crisis alerts.

SAP Hit the Mark

“As the ITZBund, we do not develop basic technologies in house,” explains Lehmann. “We buy and operate marketable solutions and continually adapt them to the needs of our customers, the federal authorities. For many years we have been working intensively and productively with SAP.” The project that the ITZBund and SAP worked on got a lot of attention. The solution offered guaranteed data protection and IT security, which is of the utmost importance when selecting the right IT solution provider. “We also deal with highly sensitive data and taxpayers’ money, among other things, and therefore we can’t afford any mishaps,” says Lehmann. “The technology has to be mature and meet our extremely stringent data protection requirements. Citizens need to trust that their data is always protected.”

AI: Platform-as-a-Service

The ITZBund and SAP joined forces and developed an AI reference architecture to be used later by various federal authorities for different projects. The experience gained from previous projects was considered when developing the platform as a service (PaaS).

The AI platform, hosted by the ITZBund data center, enables easy access for all federal authorities. The ITZBund will also take care of maintenance and support, and it plans to offer PaaS to all federal authorities who are interested in using AI.

Modular and Scalable AI Platform

“It is important for us to create a platform that we only have to invest in once but can deploy repeatedly for our customers,” says Wiebke Bedenknecht, head of Unit Software Development. “We want to create a kind of modular system that allows us to implement different technological requirements at a very high speed.” The platform supports agile development and is scalable to accommodate even more applications.

Project Timeline

Nine months after the project kicked off, the AI platform was completely functional. The first agency to go live, the German Federal Central Tax Office (BZSt), can now use AI to analyze, link, and evaluate tax data as part of its “Answer” project. With the help of clustering and natural language processing (NLP), the intelligent system detects anomalies and correlations in tax data that would otherwise be almost impossible to catch on a case-by-case basis. And another customer is already in the starting blocks: the German Federal Ministry of Finance (BMF) is scheduled to connect to the AI platform and take advantage of its full range of capabilities.

Collaborating for Germany’s Digitalization

In the future, the ITZBund will continue to rely on partnerships with the private sector; the consolidation of the federal administration is not yet complete. “As a public administration, we must have control and data sovereignty – but not do everything ourselves,” says Lehmann. “If we want digital sovereignty, we need powerful partners who will work with us in representing the interests of the German Federal Republic.”