Kreis Borken's Agentic-AI Pilot Cuts Long-Term-Care Application Processing by a Fifth

Kreis Borken, a district administration in North Rhine-Westphalia, was selected as one of 20 pilot projects (from nearly 600 applications nationwide) under the German Federal Ministry for Digital and State Modernisation's (BMDS) 'Agentic AI Hub' programme, which paired 19 municipalities with 9 AI start-ups between March and May 2026.

The pilot aimed to test whether an autonomous AI agent could pre-review 100-plus-page applications for 'Hilfe zur Pflege' (state long-term-care assistance), checking completeness and preparing financial data for means-testing while keeping a human reviewer in the loop.

Working with Berlin-based AI start-up deepset on its Haystack framework, Kreis Borken's Social Services department deployed the agent to check submissions for completeness, extract and prepare financial data, and draft a recommendation for the caseworker's decision.

Over the three-month pilot, Kreis Borken measured a 35% reduction in time spent on selected work steps and roughly a 20% reduction across the whole application process; in 15 of 18 evaluable cases, staff reported feeling 'secure' relying on the tool's output. Programme-wide, the BMDS reported that 70% of the 20 piloted solutions reached operational use, 95% of participating municipalities were highly satisfied, and half were assessed as directly transferable to other municipalities.

Results come from a short, small-sample pilot self-reported by the district and sponsoring ministry, without independent external audit; the federal programme's next step is to build a legally compliant procurement pathway before any wider rollout.