Agentic AI & GenAIDeep Dive120min
Why Your RAG Agent Is Confidently Wrong: Retrieval Choices That Actually Matter
The talk argues that retrieval, not embeddings, is the key RAG design choice. It compares vector, graph, and BM25 methods, and shows how hybrid approaches like Vector+Graph or Vector+BM25 can improve grounding and precision. It offers a practical framework for choosing retrieval by data shape and answer needs.
talk.summaryAiDisclaimer
David vonThenenNetApp
Most RAG discussions start and end with vector embeddings. That makes sense because vector search is approachable, fast to prototype, and widely supported. But semantic similarity is not the same thing as answer retrieval. When teams rely on embeddings as the default for every use case, they often end up with systems that sound convincing while returning weak, incomplete, or confidently incorrect answers. This talk reframes retrieval as the real design decision in RAG, not a backend detail.
We will walk through the major retrieval options at a high level, including vector, graph, and BM25 approaches, and explain where each one fits. Then we will show why hybrid designs, such as Vector + Graph and Vector + BM25, often produce stronger results by combining semantic context with stronger grounding and greater precision. The goal is to give AI engineers a practical mental model for choosing a retrieval approach based on the shape of their data and the kinds of answers they need, rather than defaulting to embeddings because everyone else did.
We will walk through the major retrieval options at a high level, including vector, graph, and BM25 approaches, and explain where each one fits. Then we will show why hybrid designs, such as Vector + Graph and Vector + BM25, often produce stronger results by combining semantic context with stronger grounding and greater precision. The goal is to give AI engineers a practical mental model for choosing a retrieval approach based on the shape of their data and the kinds of answers they need, rather than defaulting to embeddings because everyone else did.
David vonThenen
David vonThenen is an AI/ML Engineer where he focuses on production AI systems, enterprise AI strategy, and architectures for explainable, governable, and reliable generative AI. His work spans Agentic AI, Graph RAG, Document RAG, AI memory systems, multi-agent architectures, OpenSearch, Neo4j, data lineage, provenance, and AI governance.
David has more than 20 years of experience building production software across AI/ML, speech and NLP, Kubernetes, cloud-native platforms, storage, virtualization, and backup/recovery. He combines hands-on engineering with technical strategy, open-source leadership, developer advocacy, and customer-facing architecture work.
David has more than 20 years of experience building production software across AI/ML, speech and NLP, Kubernetes, cloud-native platforms, storage, virtualization, and backup/recovery. He combines hands-on engineering with technical strategy, open-source leadership, developer advocacy, and customer-facing architecture work.