Large-scale infrastructure permit procedures in Germany are among the most documentation-intensive processes. Reviews sometimes take longer than the projects they are meant to authorize, delaying investments despite available mandates and funding. This erodes planning reliability, pushes strategic objectives beyond legislative cycles and reduces the effective use of public resources.
To address this, the BMDS developed a solution allowing permit submissions to be screened on the same day they arrive. The system analyses application materials, structures submissions, maps case-specific facts against applicable legal norms and consolidates relevant provisions across interconnected regulatory domains. On this basis, it can generate drafts for legally grounded permit decisions under human oversight to support the review process. At this stage, the system has limited autonomy, allowing for gradual expansion once confidence and reliability increase. This reflects that scaling is not only a matter of technical feasibility – it also requires people to trust and confidently work with agentic systems.
While agentic approaches to automating legal reasoning are emerging across several fields, scaling to a truly agnostic solution is challenging due to computing capacity, cost and energy constraints. Therefore, rather than scanning the full body of administrative law, the system narrows the legal search to the relevant domains for each case – for example, a wind turbine permit only draws on specific environmental and regional regulations. This targeted approach enables scalable deployment while applying AI capacity only where it genuinely adds value.