AI - Vector Context Remove
Type identifier:
ai:vectorContext:removeCategory: AI Operations
Description
The AI - Vector Context Remove node deletes embeddings from a vector context. It can remove specific embeddings by ID or remove all embeddings matching certain label criteria.
Input Handles
Handle | Type | Description |
|---|---|---|
| control | Triggers the removal operation. |
|
| ID of the specific embedding to remove. |
|
| Dynamic handles for label-based removal. |
Output Handles
Handle | Type | Description |
|---|---|---|
| control | Triggered when removal completes. |
| control | Triggered when removal fails. |
Configuration Options
Option | Type | Default | Description |
|---|---|---|---|
Labels |
|
| Label keys for filtered removal. |
Behaviour
- Receives removal criteria (ID, labels, or both)
- Identifies embeddings matching the criteria
- Removes matching embeddings from vector context
- Triggers success or error control flow
Removal Modes
By ID:
- Removes the specific embedding with the given ID
- Exact match required
By Labels:
- When labels are configured, removes embeddings matching label values
- All specified labels must match
Use Cases
- Delete outdated document embeddings when content is updated
- Clear embeddings for a specific category or source
- Remove test data from production vector context
Examples
Remove Specific Embedding
Input:
id: "doc-123-chunk-5"
Removes the single embedding with that ID.
Remove All Document Chunks
Configuration:
- Labels:
["documentId"]
Input:
documentId: "doc-123"
Removes all embeddings with that document ID label.
Re-index Document Flow
Flow:
- AI - Vector Context Remove (by documentId)
- Text - Recursive Split (new content)
- List - Map with AI - Vector Context Add
Use case: Update indexed content when a document changes.
Cleanup by Category
Configuration:
- Labels:
["category", "status"]
Use case: Remove all "draft" content from a specific category.