This 2024 case study describes how a Fortune 500 biopharma enterprise with 34,000+ employees reduced regulatory information turnaround time by replacing a request-based process that took over 72 hours with a GenAI conversational Q&A bot that delivers answers in under 30 seconds. Brillio built separate pipelines for unstructured and structured data: unstructured queries use OpenSearch as a retriever with an Azure OpenAI generator (text-davinci-003), while structured queries use a two-stage Azure OpenAI approach (gpt-35-turbo) that converts questions to SQL and then generates a plain-language response. A classifier routes queries (structured, unstructured, follow-up, or mixed), with LangChain memory and fallback handling plus controlled access via security groups and an MS Teams UI, delivere

Join for free to read