Search that understands the meaning and intent behind a query, not just keyword matching.
Semantic search is a search methodology that understands the meaning, intent, and contextual relationships behind a query, rather than simply matching keywords. Traditional keyword search looks for exact or near-exact word matches. Semantic search uses natural language processing and vector representations to understand that 'car' and 'automobile' are synonymous, that 'best running shoes for flat feet' implies a specific need, and that the context of a document matters as much as the words it contains.
Semantic search is now the foundation of enterprise AI assistants, RAG systems, and intelligent document retrieval. As organizations deploy AI that must search across millions of internal documents, emails, and databases, keyword search fails catastrophically. Semantic search enables AI agents to find the most relevant information even when the exact query terms don't appear in the source document.
Modern semantic search works by converting text into dense vector representations (embeddings) using transformer models. Both the query and all documents in the corpus are converted to vectors in the same high-dimensional space. Search then becomes a nearest-neighbor problem: find the documents whose vectors are closest to the query vector. This is typically implemented using vector databases like Pinecone, Weaviate, or pgvector.
An enterprise deploys a semantic search system over their internal knowledge base. An employee asks 'How do I handle a customer who wants to cancel their subscription?' The system returns the relevant retention policy document even though it uses terms like 'churn prevention' and 'subscription cancellation procedures' rather than the exact words in the query.