SPARK Workflow

Project Objectives and Core Functionalities

Over the past nine months, the Federal Ministry for Digitalisation and State Modernisation has commissioned the development of AI modules as part of an innovation partnership within the SPARK project. These AI modules provide operational support for complex administrative planning and approval procedures. The development was funded by the Climate and Transformation Fund of the Federal Ministry for Economic Affairs and Energy. The AI helps to speed up administrative procedures and effectively relieve public sector staff of time-consuming tasks. The solution is designed to be agnostic in terms of procedures and areas of law, enabling broad reusability in the procedures of the federal, state and municipal authorities. At the same time, the AI modules can also be used for other tasks beyond planning and approval procedures.

Planning and approval procedures are considered a major bottleneck for the maintenance and expansion of infrastructure in Germany. The underlying administrative processes are often complex, resource-intensive and time-consuming. SPARK supports staff in public authorities through a workflow tailored to their legal working methods, as well as through AI modules that provide them with initial review and assessment suggestions. Users always make the decisions during the review process. AI suggestions can be adopted, supplemented or rejected entirely.

This has resulted in several modules and services that map out key procedural steps. These include, in particular, the extraction of factual content from application documents, formal and substantive completeness checks including plausibility checks, public and stakeholder engagement, legal review and assessment, and the drafting of decisions. The AI solution is complemented by a digital legal doctrine system featuring legal databases, automated analysis of legal provisions and legal assessment mechanisms.

Project Status and Open-Source Release

SPARK is being released as open source in several phases. The solution is currently in the beta stage.

The first release includes the content extraction and formal completeness checking modules, including plausibility checks, as well as supporting backend services. Further functions will be added in subsequent open-source releases.

A detailed overview of the contents and timing of all three releases is provided in roadmap.md."