AI Search reranker

An optional second pass on AI Search queries: after ANN, hybrid or full-text retrieval, a Databricks cross-encoder model re-scores the results against the query and returns them in a better relevance order. You turn it on per query and name the columns to rerank, and the model reads only the first 2,000 characters of those columns. Databricks says it typically improves retrieval quality by about 10 percent and usually adds under a second of latency per query. It runs as a Databricks Designated Service using Databricks Geos, so workspaces outside the US, EU, Australia and New Zealand may need cross-Geo processing turned on. It launched as the Mosaic AI Vector Search reranker.

Status and dates are extracted from public sources only. We don't see the private AE-shared roadmap. If something looks wrong, tell us.