Glossary
Vector search, explained
Vector search is a technique that finds content by meaning, comparing embeddings of the question and of the documents, instead of looking only for exact words.
It can surface the return policy when a customer asks "can I send back what I bought", even if the document never uses those words.
Many systems combine vector search with keyword search, known as hybrid search, so they do not miss codes, product names and acronyms.
In RAG, vector search decides which passages reach the model. If it retrieves the wrong passage, the answer is wrong even with a good LLM.
Conversations aren't support.They're infrastructure.
Step into the new era of the relationship between companies and customers.