From Transkribus to Database. Post-processing Parish Registers into Structured Data

After Transkribus transcribes a 19th-century parish register, what comes next? This talk reveals a practical multi-LLM post-processing pipeline — combining ChatGPT, Claude, and Gemini — to clean, normalise, and structure raw transcriptions from multilingual Romanian, Hungarian, Latin, and German records into relational datasets ready for historical and genealogical research.