Document Processing · Enterprise Knowledge Search
Caseworkers in Landratsamt Augsburg's Youth Welfare Office had to manually research legal references, citizen-submitted documents, and other records before working a case, making the process slow, labor intensive, and prone to errors during a staffing shortage.
Reported impact
The pilot materially reduced case-compilation time, improved the quality of generated summaries, and showed Landratsamt Augsburg how gen AI could free caseworkers to focus more on citizens and less on repetitive research work.
Landratsamt Augsburg teamed with IBM Client Engineering and IBM Consulting to prototype a generative-AI case-research assistant for youth welfare staff. After workshops and a three-day hackathon, the joint team built a customized web application integrated with IBM watsonx Assistant, IBM watsonx.ai, and IBM Watson Discovery. The application queries multiple disparate sources defined by the client, retrieves supporting material, and summarizes findings for each welfare case so staff can compile the legal and financial context faster. Watson Discovery provides retrieval across the various references and submitted documents, watsonx.ai performs summarization, and watsonx Assistant supplies the conversational interface for the web app. The pilot showed that a workflow that could take up to an hour to compile manually could be reduced to a few minutes, giving the office a safer path to using gen AI in a sensitive public-sector process.
Legal references, citizen-submitted documents, and supplementary records are queried through Watson Discovery, the retrieved material is summarized by watsonx.ai, and caseworkers interact with the results through a watsonx Assistant-powered web application for each welfare case.
During the three-day event, 8-10 engineers came together to build a customized web-application that was integrated with IBM watsonx Assistant and supported by IBM watsonx.ai and IBM Watson Discovery.
It was designed to query multiple disparate data sources set forth by the client and summarize findings for each case.
The tests conducted during the event demonstrated an up to 91% improvement in the time required to compile case data.
before the pilot, it could take up to one hour ... whereas during the pilot ... as few as five minutes
these individuals conduct extensive research from a variety of different data sources, such as online legal references, documents submitted by citizens and other supplementary documentation
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
IBM Sample Feb 25, 2026 General State Attorney’s Office The General State Attorney’s Office needed to manage and access a very large volume of legal information more efficiently so legal professionals could find relevant precedents, compare lawsuits, and prepare responses… IBM cloud · IBM watsonx · TEMIS 
OpenAI Sample Aug 31, 2026 Polimill Polimill needed to reduce administrative burden across fragmented Japanese municipal systems so public employees could find prior decisions, prepare assembly responses, and deliver services without each municipality… 3-5x development speed increase from Codex and OpenAI hands-on support Codex · GPT models · OpenAI API 
NVIDIA Sample Jul 7, 2026 Public Ministry of Rio Grande do Sul (MPRS) MPRS needed to manage rising legal caseloads, archived police inquiries, and public-service requests more efficiently, but the volume of legal documents and the need for compliant, locally controlled analysis made… 497 municipalities Archiving Intelligence Assistant (AIA) · Citizen Access Agent · Justice Intelligence platform 