Private Embeddings & Vector Search.

What Are Embeddings?

An embedding is a list of numbers -- a vector -- that represents the meaning of a piece of text. The word "dog" and the phrase "canine companion" have very different spellings but similar embeddings because they mean similar things. This is the foundation of semantic search: finding documents by meaning, not just keyword matching.

Traditional search (like grep or Ctrl+F) finds exact text matches. Semantic search with embeddings finds conceptually related content. Search for "employee burnout" and find documents about "staff retention challenges" and "workplace wellness programs" even if they never use the word "burnout."

When embeddings are generated by a cloud API, every document you embed is sent to a third-party server. With local embeddings, the model runs on your machine -- your documents never leave.

Generating Local Embeddings

Ollama makes local embedding generation trivial:

# Pull an embedding model
ollama pull nomic-embed-text

# Generate an embedding via API
curl http://localhost:11434/api/embed -d '{
  "model": "nomic-embed-text",
  "input": "This is my private document content."
}'

The response includes a vector (array of 768 floating-point numbers for nomic-embed-text). This vector is a mathematical fingerprint of that text's meaning.

In Python:

import requests

def get_embedding(text):
    response = requests.post(
        "http://localhost:11434/api/embed",
        json={"model": "nomic-embed-text", "input": text}
    )
    return response.json()["embeddings"][0]

# Embed a document
vector = get_embedding("Quarterly revenue increased 15% year over year.")
print(f"Vector dimension: {len(vector)}")  # 768