Commoditizing Inference - Your Query Language Should Speak AI

Explore how AI inference is being integrated as a native primitive within modern search stacks through this 46-minute conference talk from Haystack EU 2025. Learn about commoditizing inference capabilities like semantic retrieval, result re-ranking, and text generation directly within query languages, treating them as fundamental operations alongside traditional filters and scoring functions. Discover Elasticsearch's inference APIs and understand how inference can be invoked directly from the query DSL, examining the benefits of query-layer integration including simplicity, composability, and accessibility while considering the tradeoffs compared to application-level inference management. Gain practical insights into new inference primitives in Elasticsearch Query Language (ES|QL) including EXT_EMBEDDING, COMPLETION, and RERANK functions that bridge the gap between low-level code control and declarative DSL expressiveness. Follow along with practical, example-driven demonstrations designed for search practitioners and analysts looking to leverage AI capabilities within their query workflows.