The Components Service API

These are three calls allowing you to check the status of the app, perform a semantic search and generate an embedding.

Health check (no authentication required

curl "https://api.example.com/components/__gtg"

Perform a semantic search

curl -X POST "https://api.example.com/components/vector/search?text=What%20is%20GraphRAG&top_k=3"

Generate an embedding

curl -X POST "https://api.example.com/components/vector/embed?text=Hello%20world"

Embeddings

An embedding is a numerical vector representation of text. The Components Service uses a configured embedding provider (AWS Bedrock or OpenAI) to convert your text into a vector that captures semantic meaning.

"GraphRAG combines knowledge graphs with RAG"
                    ↓
[-0.0116, -0.0209, -0.0105, 0.0075, -0.0234, ... ] (vector of floats)

Vector search

Vector search finds documents with similar meaning to your query, not just matching keywords. Your query is converted into an embedding and then compared with all stored document embeddings using cosine similarity.

Query: "machine learning applications"
                    ↓
Finds documents about: AI, neural networks, deep learning, data science(
even if they do not contain "machine learning")

Similarity scores

The results include a score ranging between 0 and 1.

The Components Service does not require API keys or tokens for requests.