Semantic Space

Explore how AI understands meaning through vector embeddings

What you're seeing

Each concept is converted into a 1024-dimensional vector — a point in "meaning space". Cooking instructions cluster together. Programming concepts cluster together. Sports descriptions cluster together. Not by keywords, but by meaning.

0 Concepts

← Select a concept from the list

See which concepts the AI considers semantically similar — sharing meaning, not just words.

How it works:

Each concept becomes a 1024-dimensional vector — a point in meaning-space.

Similar meanings → nearby vectors → high similarity %

Embedding Model

BGE-large-en-v1.5 (1024 dims)

Vector Store

Upstash Vector (serverless)

Similarity Metric

Cosine distance (0-100%)

"Embeddings turn meaning into geometry" — this is how RAG, semantic search, and AI memory work.