Semantic search
A search approach that ranks results by meaning rather than exact keyword overlap, using embeddings to match a query to conceptually similar content.
In more detail
Keyword search matches strings; semantic search matches meaning. By comparing the embedding of a query to the embeddings of stored content, it surfaces material that is relevant even when it uses different words than the query. A search for cancel my subscription can find a document about ending your plan.
Semantic search rarely replaces keyword search outright; the two are often combined, since exact-match still matters for names, codes, and precise terms. But for open-ended questions over documents and knowledge, matching by meaning is what makes retrieval useful for downstream AI.
Where this shows up at Ceven
Ceven leans on semantic retrieval so research and knowledge steps find the right material by meaning, not just by keyword. When a workflow needs to answer from a body of documents or gathered sources, semantic search surfaces the relevant passages, which the model then reasons over to produce a grounded result.