Quick Start

Get up and running in 5 minutes.

1. Set Up Credentials

In n8n, create Postgres credentials:

  1. Go to Credentials → Add Credential
  2. Search for Postgres
  3. Fill in your connection details:
Field Value
Host Your PostgreSQL host
Port 5432 (default)
Database Your database name
User Database user
Password Database password
SSL Configure as needed

2. Initialize Schema

Before storing embeddings, initialize the database schema.

Add a PGVector Advanced node:

Setting Value
Operation Admin
Admin Operation Ensure Schema
Dimensions 1536 (or your embedding size)

This creates the embeddings table and necessary indexes if they don’t exist.

3. Store Your First Embedding

Add another PGVector Advanced node:

Setting Value
Operation Upsert
Collection my_documents
External ID doc-001
Content “This is my first document”
Embedding [0.1, 0.2, 0.3, …]
Metadata {“category”: “test”}

Tip: Use an OpenAI Embeddings node or similar to generate the embedding vector from your text.

4. Query Similar Documents

Add a PGVector Advanced node to search:

Setting Value
Operation Query
Collection my_documents
Query Embedding [0.1, 0.2, …]
Top K 10
Distance Metric cosine

Results include:

5. Complete Example Workflow

Here’s a typical semantic search workflow:

[HTTP Request] → [OpenAI Embeddings] → [PGVector Upsert]
                         ↓
              [Search Query Input]
                         ↓
              [OpenAI Embeddings] → [PGVector Query] → [Results]

Workflow Steps:

  1. Ingest documents
    • Fetch documents (HTTP, file, database)
    • Generate embeddings with OpenAI/Cohere
    • Upsert to PGVector with metadata
  2. Search
    • Receive search query
    • Generate query embedding
    • Query PGVector for similar documents
    • Return results

Common Patterns

Upsert with External ID (Sync-Friendly)

{
  "operation": "upsert",
  "collection": "documents",
  "externalId": "notion-page-abc123",
  "content": "Page content here",
  "metadata": {
    "source": "notion",
    "lastUpdated": "2024-01-15"
  },
  "embedding": [...]
}

Using externalId means:

Batch Upsert

For bulk operations, use batch mode with field mapping:

{
  "operation": "upsert",
  "mode": "batch",
  "collection": "documents",
  "fieldMapping": {
    "externalIdField": "id",
    "contentField": "text",
    "metadataField": "meta",
    "embeddingField": "vector"
  }
}

Input items:

[
  {"id": "doc-1", "text": "First doc", "meta": {}, "vector": [...]},
  {"id": "doc-2", "text": "Second doc", "meta": {}, "vector": [...]}
]

Query with Metadata Filter

{
  "operation": "query",
  "collection": "documents",
  "queryEmbedding": [...],
  "topK": 5,
  "metadataFilter": {
    "category": "technology",
    "status": "published"
  }
}

Next Steps