Insert points

Add individual points or batches of points to a collection.

Insert individual points or batches of points into your collection. Each point can carry a JSON payload with metadata fields such as categories, prices, and timestamps. Payload fields are available for filtering during search. The payload field accepts any valid JSON structure. Include only the fields you need for filtering or displaying results.

You need an existing collection. Vectors must match the dimension configured for the collection. See Create a collection to set one up.

Insert a single point

The insert() method is an alias for upsert(). A new ID inserts a new point, while an existing ID updates the point with the new vector and payload.

Python

import random
from actian_vectorai import VectorAIClient, PointStruct

DIMENSION = 128
COLLECTION = "products"

# Connect to VectorAI DB server
with VectorAIClient("localhost:6574") as client:
    # Generate vector from your embedding model
    vector = [random.gauss(0, 1) for _ in range(DIMENSION)]
    
    # Insert point with payload
    point = PointStruct(
        id=1,  # Point ID
        vector=vector,  # Vector embedding
        payload={  # Metadata (optional)
            "name": "Laptop",
            "category": "electronics",
            "price": 999.99,
            "in_stock": True
        }
    )
    
    # Upsert point to collection
    client.points.upsert(COLLECTION, [point])
    print("Point inserted successfully")

Payload is optional. You can insert points with only ID and vector:

point = PointStruct(id=1, vector=vector)  # No payload
client.points.upsert(COLLECTION, [point])

Batch insert points

Batch operations are significantly faster than individual inserts. Use batch sizes between one hundred and one thousand points for optimal performance. The upsert_points() method is an alias for batch_upsert().

Python

import random
from actian_vectorai import VectorAIClient, PointStruct

DIMENSION = 128
COLLECTION = "products"

# Connect to VectorAI DB server
with VectorAIClient("localhost:6574") as client:
    random.seed(42)  # Reproducible random vectors
    
    # Prepare batch data
    products = [
        {"name": "Smartphone", "category": "electronics", "price": 699.99, "in_stock": True},
        {"name": "Tablet", "category": "electronics", "price": 499.99, "in_stock": True},
        {"name": "T-Shirt", "category": "clothing", "price": 29.99, "in_stock": True},
        {"name": "Jeans", "category": "clothing", "price": 79.99, "in_stock": False},
        {"name": "Coffee Beans", "category": "food", "price": 12.99, "in_stock": True},
        {"name": "Energy Bar", "category": "food", "price": 2.99, "in_stock": True},
        {"name": "Python Book", "category": "books", "price": 39.99, "in_stock": False},
        {"name": "AI Textbook", "category": "books", "price": 89.99, "in_stock": True},
        {"name": "Action Figure", "category": "toys", "price": 24.99, "in_stock": True},
        {"name": "Board Game", "category": "toys", "price": 49.99, "in_stock": False}
    ]
    
    # Create points with vectors and payloads
    points = [
        PointStruct(
            id=i + 1,  # Point ID
            vector=[random.gauss(0, 1) for _ in range(DIMENSION)],  # Generate vector
            payload=payload  # Attach metadata (optional)
        )
        for i, payload in enumerate(products)
    ]
    
    # Batch insert all points
    client.points.upsert(COLLECTION, points)
    
    print(f"Successfully inserted {len(points)} points")