Graph Index: Search vs. Semantic Retrieval

Both approaches can retrieve content from the same Graph-indexed content. The difference is how we want to find that content and what experience we're building.

Search Query — "Find what I typed"

Use Search when the employee is trying to locate a specific document, page, person, or piece of content.

Best for:

Traditional enterprise search
Known documents or titles
Keyword searches
Search results with filters, sorting, etc.

Example

A document says:

"Our café serves hamburgers with cheese, lettuce, tomato, and fries."

Employee searches:

"Cheeseburger"

Traditional search may struggle because "cheeseburger" doesn't actually appear in the document.

Semantic Retrieval — "Find what I mean"

Use Semantic Retrieval when the employee is asking a question or describing what they need, and you want to find content based on meaning rather than exact words.

Best for:

AI-generated answers
AI Chat / RAG
Natural-language questions
Situations where employees don't know the terminology used in the content

Example

The same document says:

"Our café serves hamburgers with cheese, lettuce, tomato, and fries."

Employee asks about:

"Cheeseburgers"

Semantic retrieval can understand that a hamburger with cheese is conceptually a cheeseburger, even though that exact word isn't present.

Simple Guiding Principle

Search = Find the content that matches what I typed.

Semantic Retrieval = Find the content that matches what I meant.

Use Search when employees are looking for something. Use Semantic Retrieval when AI is trying to understand and answer something.