Q&A with RAG Accelerator

This accelerator enables high-accuracy question answering over customer-defined knowledge corpuses in watsonx.data (OpenSearch/Milvus/DataStax) or watsonx Discovery/Elastic, with end-user feedback collection, topic detection, and user satisfaction reports by topic with drill-down insights to improve the corpus. To build a corpus, it converts HTML, PDF, DOCX, PPT, MD, TXT to text, splits into chunks, and indexes them. It deploys a Python AI Service that queries the index, fetches top results, runs LLM inference with hallucination checks. Feedback logs and notebooks analyze user insights and performance metrics. Click Create a project to start or Log In. No account? Sign Up for free.