BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//pyconde-pydata-2025//speaker//8MXU3M
BEGIN:VTIMEZONE
TZID:Europe/Berlin
BEGIN:DAYLIGHT
DTSTART:20240424T000000
TZNAME:CEST
TZOFFSETFROM:+0200
TZOFFSETTO:+0200
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20241027T030000
RDATE:20251026T030000
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20250330T030000
RDATE:20260329T030000
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:Lessons learned in bringing a RAG chatbot with access to 50k+ dive
 rse documents to production - Bernhard Schäfer\, Nico Mohr
DTSTART;TZID=Europe/Berlin:20250424T165500
DTEND;TZID=Europe/Berlin:20250424T174000
DTSTAMP:20260809T110748Z
UID:pretalx-pyconde-pydata-2025-XLZQFA@pretalx.com
DESCRIPTION:Retrieval-Augmented Generation (RAG) chatbots are a key use ca
 se of GenAI in organizations\, allowing users to conveniently access and q
 uery internal company data. A first RAG prototype can often be created in 
 a matter of days. But why are the majority of prototypes still in the pilo
 t stage? [\\[1\\]](https://www2.deloitte.com/content/dam/Deloitte/us/Docum
 ents/consulting/us-state-of-gen-ai-q3.pdf)\n\nIn this talk we share our in
 sights from developing a production-grade chatbot at Merck. Our RAG chatbo
 t for R&D experts accesses over 50\,000 documents across numerous SharePoi
 nt sites and other sources. We identified three technical key success fact
 ors:\n1. Building a robust data pipeline that syncs documents from source 
 systems and that handles enterprise features such as replicating user perm
 issions. \n2. Developing a chatbot workflow from user question to answer w
 ith retrieval components such as hybrid search and reranking\n3. Establish
 ing a comprehensive evaluation framework with a clear optimization metric.
 \n\nWe think that many of these lessons are broadly applicable to RAG chat
 bots\, making this talk valuable for practitioners aiming to implement Gen
 AI solutions in business contexts.
LOCATION:Titanium3
URL:https://pretalx.com/pyconde-pydata-2025/talk/XLZQFA/
END:VEVENT
END:VCALENDAR