Semantic Search

Traditional keyword search matches exact words — searching for "pricing" only finds entries containing the word "pricing." Semantic search is fundamentally different. It understands meaning. When you search for "how much do we charge," it finds entries about pricing, rates, fees, and costs — even if none of them contain the exact phrase you searched for. This works through vector embeddings, mathematical representations of meaning that allow the system to measure how semantically similar two pieces of text are.

Sarudo uses a hybrid search approach that combines semantic search with traditional keyword matching. This gives you the best of both worlds — the meaning-understanding capability of semantic search and the precision of keyword matching. When you ask a question, the system runs both types of search in parallel and combines the results, prioritizing entries that score high on both semantic relevance and keyword match. This produces more accurate results than either approach alone.

Semantic search sits behind most of what your AI employee does. When you ask a question, it searches your knowledge base for relevant information. When it drafts an email, it searches for the contact's details and recent interactions. When it prepares a proposal, it searches for your pricing and template preferences. You rarely need to explicitly "search" — it happens as part of handling your request.

The quality of search results improves as your knowledge base grows and becomes more organized. You can improve search quality by being specific when storing knowledge (include context and details), using consistent terminology across your knowledge entries, categorizing knowledge appropriately, and periodically reviewing and updating outdated information. The more structured and detailed your knowledge base, the more accurate and useful search results will be.