Definition
Embeddings place similar concepts near each other in a mathematical space. They are commonly used to find relevant documents even when the wording does not exactly match a search query.
Why it matters
They make semantic search and RAG possible, helping AI retrieve meaning rather than only keywords.
Business example
A policy assistant finds the right benefits document when an employee asks about parental leave in their own words.
When to use it
Use them when people search with varied language across a meaningful body of content.
When not to use it
Do not use them as a substitute for structured filters when an exact field such as an invoice number is available.
How Automathing approaches it
We combine semantic retrieval with metadata filters and source governance, rather than treating embeddings as magic search.