Landratsamt Augsburg (Youth Welfare Office)

Caseworkers in the Youth Welfare Office at Landratsamt Augsburg must determine the appropriate cost contribution for each youth welfare, integration assistance, and childcare case under their purview. Doing so requires extensive manual research across online legal references, citizen-submitted documents, and other supplementary sources before work on a case can even begin. The process is thorough but time-consuming and error-prone, compounded by a shortage of specialized research staff.

Landratsamt Augsburg partnered with IBM Client Engineering and IBM Consulting, running workshops to pinpoint pain points before testing generative AI in a three-day hackathon. A team of 8-10 engineers built a custom web application integrated with IBM watsonx Assistant and supported by IBM watsonx.ai and IBM Watson Discovery to query multiple disparate data sources and automatically summarize the findings for each case.

Tests conducted during the hackathon pilot showed up to a 91% improvement in the time required to compile case data, cutting the process from as long as one hour down to as little as five minutes. The team also anecdotally observed improved quality and accuracy in the generated case summaries and is now looking to expand generative AI use across the group.

• A time-boxed hackathon let a team with mixed skepticism test gen AI on sensitive casework before committing to a full rollout.
• Summarizing and querying across disparate legal references and citizen-submitted documents can sharply cut manual research time.
• Quality/accuracy gains were only observed anecdotally, not independently measured, and the initiative remains a pilot rather than a full deployment.