Update

Updates an existing vector's values or metadata in a Pinecone index. Use this when you need to modify a vector without changing its ID.

Common Properties

• Name - The custom name of the node.
• Color - The custom color of the node.
• Delay Before (sec) - Waits in seconds before executing the node.
• Delay After (sec) - Waits in seconds after executing node.
• Continue On Error - Automation will continue regardless of any error. The default value is false.

info

If the ContinueOnError property is true, no error is caught when the project is executed, even if a Catch node is used.

Inputs

• Connection Id - Connection identifier from Connect node (optional if API Key credential is provided directly).
• Host URL - Your index's host URL. Format: https://INDEX_NAME-PROJECT_ID.svc.ENVIRONMENT.pinecone.io. You can find this in your Pinecone console.
• Id - The unique ID of the vector to update. Must already exist in the index.
• Values - New vector values array (same dimension as index). The vector embedding data to update.

Options

• API Key - Pinecone API key credential (optional - use this instead of Connection Id if not using Connect node).
• Metadata - New metadata object to set for the vector. This will replace the existing metadata entirely.
• Name Space - Namespace containing the vector to update. Optional, defaults to the default namespace.

Output

• Response - Object confirming the update operation (typically an empty object {}).

How It Works

The Update node modifies an existing vector in your Pinecone index. When executed, the node:

1. Validates required inputs (Host URL, Id, Values)
2. Constructs the update request with new values and optional metadata
3. Sends a POST request to the index's /vectors/update endpoint
4. Returns a confirmation response

Requirements

• An existing Pinecone index with data
• A valid vector ID that exists in the index
• New vector values with the same dimension as the index
• The vector must exist in the specified namespace

Error Handling

The node will return specific errors in the following cases:

• ErrInvalidArg - Host URL is empty
• ErrInvalidArg - Id is empty
• ErrInvalidArg - Invalid Connection ID or missing API key
• ErrInternal - Response format is not valid
• ErrStatus - HTTP error from Pinecone API (vector not found, dimension mismatch, etc.)

Usage Notes

• The vector ID must already exist (Update won't create new vectors - use Upsert for that)
• If Values is provided, it must match the index dimension
• You can update just metadata by providing only metadata without values
• Metadata update replaces the entire metadata object, not just specified fields
• The update is atomic - either all changes succeed or none do
• Updates are eventually consistent

Update vs Upsert

Use Update when:

• You want to modify an existing vector
• You need to ensure the vector exists before modifying
• You want explicit error if vector doesn't exist

Use Upsert when:

• You want to insert or update (create if not exists)
• You're batch loading data
• You don't care if the vector already exists

Best Practices

• Verify the vector exists before updating (use Fetch)
• Keep the same dimension when updating values
• Use descriptive metadata fields
• Document why updates are happening
• Consider using Upsert for simpler logic unless you specifically need Update behavior
• Batch updates when possible (though Update is single-vector only)

Example: Update Vector Values

// Update the embedding for an existing document
const documentId = "doc_123";
const newEmbedding = await generateEmbedding(updatedDocumentText);

// Configure Update node:
msg.id = documentId;
msg.values = newEmbedding;

// After Update executes
console.log(`✓ Updated vector values for ${documentId}`);

Example: Update Metadata Only

// Update metadata without changing the vector values
const productId = "product_456";

msg.id = productId;
msg.values = null; // Don't update vector values
msg.metadata = {
  price: 29.99,
  in_stock: true,
  last_updated: new Date().toISOString(),
  category: "electronics"
};

// After Update executes
console.log(`✓ Updated metadata for ${productId}`);

Example: Update with Namespace

// Update a vector in a specific namespace
msg.id = "user_789";
msg.namespace = "user-profiles";
msg.values = newUserEmbedding;
msg.metadata = {
  name: "John Doe",
  tier: "premium",
  last_active: new Date().toISOString()
};

console.log("✓ Updated user profile in user-profiles namespace");

Example: Conditional Update Based on Fetch

// Step 1: Fetch current vector
// Execute Fetch node for ID "doc_123"

const currentVector = msg.resFetch.vectors["doc_123"];

if (!currentVector) {
  throw new Error("Vector not found - cannot update");
}

// Step 2: Modify metadata
const updatedMetadata = {
  ...currentVector.metadata,
  views: (currentVector.metadata.views || 0) + 1,
  last_viewed: new Date().toISOString()
};

// Step 3: Update with new metadata
msg.id = "doc_123";
msg.values = currentVector.values; // Keep same values
msg.metadata = updatedMetadata;

// Execute Update node
console.log("✓ Incremented view count");

Example: Refresh Embeddings for Updated Content

// When document content changes, update its embedding
const documents = [
  { id: "doc_1", text: "Updated content for document 1" },
  { id: "doc_2", text: "Updated content for document 2" }
];

for (const doc of documents) {
  // Generate new embedding
  const embedding = await generateEmbedding(doc.text);

  // Update vector
  msg.id = doc.id;
  msg.values = embedding;
  msg.metadata = {
    text: doc.text,
    updated_at: new Date().toISOString(),
    version: 2
  };

  // Execute Update node
  console.log(`✓ Updated ${doc.id}`);
}

Example: Update Status Flag

// Mark document as reviewed
const documentId = "doc_456";

// Fetch current metadata
const current = msg.resFetch.vectors[documentId];

// Update with new status
msg.id = documentId;
msg.values = current.values; // Keep vector same
msg.metadata = {
  ...current.metadata,
  reviewed: true,
  reviewed_by: "admin",
  reviewed_at: new Date().toISOString()
};

console.log("✓ Marked document as reviewed");

Example: Batch Update Pattern

// Update multiple vectors (requires loop or multiple Update nodes)
const updates = [
  { id: "vec_1", values: newEmbedding1, metadata: metadata1 },
  { id: "vec_2", values: newEmbedding2, metadata: metadata2 },
  { id: "vec_3", values: newEmbedding3, metadata: metadata3 }
];

for (const update of updates) {
  msg.id = update.id;
  msg.values = update.values;
  msg.metadata = update.metadata;

  // Execute Update node
  console.log(`✓ Updated ${update.id}`);

  // Add small delay to avoid rate limits
  await sleep(100);
}

console.log(`Updated ${updates.length} vectors`);

Example: Correct Dimension Mismatch

// If you need to update vectors with different dimensions,
// you must delete and recreate (Update won't work)

const vectorId = "vec_old_dimension";
const oldDimension = 384;
const newDimension = 1536;

// Fetch current metadata
const current = msg.resFetch.vectors[vectorId];

if (current.values.length !== newDimension) {
  console.log("Dimension mismatch - using delete and upsert");

  // Delete old vector
  msg.idsToDelete = [vectorId];
  // Execute Delete node

  // Upsert with new dimension
  msg.ids = [vectorId];
  msg.values = [newEmbeddingWithNewDimension];
  msg.metadatas = [current.metadata];
  // Execute Upsert node
} else {
  // Normal update
  msg.id = vectorId;
  msg.values = newEmbeddingWithNewDimension;
  // Execute Update node
}

Example: Merge Metadata

// Add new fields to metadata without losing existing ones
const vectorId = "product_123";

// Fetch current
const current = msg.resFetch.vectors[vectorId];

// Merge new fields
const updatedMetadata = {
  ...current.metadata,
  discount: 0.15,
  featured: true,
  promotion_end: "2024-12-31"
};

msg.id = vectorId;
msg.values = current.values;
msg.metadata = updatedMetadata;

console.log("✓ Added promotion metadata");

Example: Update with Validation

const vectorId = "doc_789";
const newEmbedding = msg.newEmbedding;
const indexDimension = 1536;

// Validate dimension
if (newEmbedding.length !== indexDimension) {
  throw new Error(
    `Dimension mismatch: Expected ${indexDimension}, got ${newEmbedding.length}`
  );
}

// Validate ID exists
const existing = msg.resFetch.vectors[vectorId];
if (!existing) {
  throw new Error(`Vector ${vectorId} not found`);
}

// Proceed with update
msg.id = vectorId;
msg.values = newEmbedding;

console.log("✓ Validation passed, updating vector");

Example: Track Update History in Metadata

const vectorId = "doc_456";

// Fetch current
const current = msg.resFetch.vectors[vectorId];

// Build update history
const updateHistory = current.metadata.update_history || [];
updateHistory.push({
  timestamp: new Date().toISOString(),
  updated_by: msg.currentUser,
  reason: "Content refresh"
});

// Keep only last 10 updates
const recentHistory = updateHistory.slice(-10);

msg.id = vectorId;
msg.values = newEmbedding;
msg.metadata = {
  ...current.metadata,
  update_history: recentHistory,
  last_updated: new Date().toISOString()
};

console.log("✓ Updated with history tracking");

Example: Selective Field Update Pattern

// Since Update replaces entire metadata, use this pattern
// to update only specific fields

const vectorId = "item_123";

// 1. Fetch current vector
const current = msg.resFetch.vectors[vectorId];

// 2. Create updated metadata with only changed fields
const updatedMetadata = {
  ...current.metadata,  // Keep all existing fields
  price: 49.99,         // Update only price
  stock: 150            // Update only stock
};

// 3. Update with merged metadata
msg.id = vectorId;
msg.values = current.values;  // Keep vector same
msg.metadata = updatedMetadata;

console.log("✓ Updated specific fields while preserving others");

Troubleshooting

Error: "Id cannot be empty"

• Ensure Id input is provided
• Check that the variable reference is correct
• Verify the ID string is not empty

Error: "Vector not found"

• The vector with this ID doesn't exist
• Check the ID for typos (case-sensitive)
• Verify you're in the correct namespace
• Use Fetch to verify the vector exists before updating

Error: "Dimension mismatch"

• New vector values have wrong dimension
• Ensure values array length matches index dimension
• You cannot change dimension with Update (need delete + upsert)

Update seems to have no effect

• Updates are eventually consistent, wait a moment
• Verify you're updating the correct namespace
• Check that the Update executed without errors
• Use Fetch to verify the update took effect

Metadata not updating as expected

• Update replaces entire metadata object
• You must include all existing fields you want to keep
• Use Fetch first to get current metadata, then merge

Cannot update vectors in specific namespace

• Verify the namespace parameter is set correctly
• Namespace names are case-sensitive
• The vector must exist in that namespace

Values parameter issues

• Values must be an array of numbers (floats)
• Must match exact dimension of index
• Cannot be null if provided (omit the parameter instead)