Re-ingest Knowledge Item

Re-runs the ingestion pipeline (extraction, chunking, embedding, indexing) for an existing knowledge item. Useful when the source content has changed or if the initial ingestion failed.

HTTP Request

$
POST https://app.resemble.ai/api/v2/knowledge_items/{uuid}/reingest

URL Parameters

Example Request

$
curl --request POST "https://app.resemble.ai/api/v2/knowledge_items/550e8400-e29b-41d4-a716-446655440000/reingest" \
>
  -H "Authorization: Bearer YOUR_API_TOKEN" \
>
  -H "Content-Type: application/json"

Response (Success)

1
{
2
  "success": true,
3
  "item": {
4
    "uuid": "550e8400-e29b-41d4-a716-446655440000",
5
    "title": "Product FAQ",
6
    "description": "FAQ content",
7
    "source_type": "url",
8
    "status": "pending",
9
    "error_message": null,
10
    "chunk_count": null,
11
    "url": "https://example.com/faq",
12
    "agent_count": 2,
13
    "created_at": "2025-12-01T10:00:00Z",
14
    "updated_at": "2025-12-01T12:00:00Z"
15
  },
16
  "message": "Re-ingestion has been queued"
17
}

Error Response (Already Processing)

1
{
2
  "success": false,
3
  "errors": {
4
    "knowledge_item": [{ "error": "already_processing", "message": "Knowledge item is already being processed" }]
5
  },
6
  "message": "Failed to reingest knowledge item"
7
}

Error Response (Rate Limited)

1
{
2
  "success": false,
3
  "errors": {
4
    "knowledge_item": [{ "error": "rate_limited", "message": "Must wait 1 minute between reingest requests" }]
5
  },
6
  "message": "Failed to reingest knowledge item"
7
}

Constraints

Rate Limiting

• Must wait 1 minute between re-ingest requests for the same knowledge item
• This prevents excessive processing and ensures system stability
• The rate limit counter resets after successful completion or failure

Status Requirements

• Cannot re-ingest items that are currently processing
• Can re-ingest items with status: pending, ready, or failed
• If status is processing, wait until it completes before attempting re-ingest

Re-ingestion Process

When you trigger re-ingestion:

1. Status Reset: Knowledge item status changes to pending
2. Chunk Cleanup: Existing chunks in search index are marked for removal
3. Queue Job: New ingestion job is queued
4. Processing Begins: Status transitions to processing
5. Fresh Extraction:
   • For URLs: Fetches latest content from the URL
   • For Documents: Re-processes the uploaded PDF file
   • For Text: Re-processes the stored text
6. Re-chunking: Content is chunked again (may produce different chunks)
7. Re-embedding: New vector embeddings are generated
8. Re-indexing: New chunks replace old ones in search index
9. Completion: Status becomes ready or failed

Use Cases

Content Updates

When the source content has changed:

$
# Website FAQ has been updated with new information
>
curl --request POST "https://app.resemble.ai/api/v2/knowledge_items/faq-uuid/reingest" \
>
  -H "Authorization: Bearer YOUR_API_TOKEN"

Failed Ingestion

Retry after fixing issues that caused initial failure:

$
# Original URL was temporarily down, now it's accessible
>
curl --request POST "https://app.resemble.ai/api/v2/knowledge_items/failed-uuid/reingest" \
>
  -H "Authorization: Bearer YOUR_API_TOKEN"

Improved Processing

Take advantage of improved ingestion algorithms:

$
# Platform has improved chunking/embedding, re-ingest for better results
>
curl --request POST "https://app.resemble.ai/api/v2/knowledge_items/old-uuid/reingest" \
>
  -H "Authorization: Bearer YOUR_API_TOKEN"

Monitoring Progress

After triggering re-ingestion, poll the Get Knowledge Item endpoint to monitor progress:

$
# Check status every few seconds
>
curl --request GET "https://app.resemble.ai/api/v2/knowledge_items/550e8400-e29b-41d4-a716-446655440000" \
>
  -H "Authorization: Bearer YOUR_API_TOKEN"

Status progression:

• pending → processing → ready (success)
• pending → processing → failed (error, check error_message)

Impact on Agents

During Re-ingestion

• Agents that have this knowledge item attached will continue to use the old chunks during re-ingestion
• No interruption to agent functionality
• RAG retrieval uses the existing chunks until new ones are ready

After Re-ingestion

• Once status becomes ready, agents automatically start using the new chunks
• No need to re-attach the knowledge item to agents
• Immediate availability of updated knowledge

Best Practices

Scheduling Re-ingestion

• URL Sources: Re-ingest periodically (daily, weekly) to keep content fresh
• Document Sources: Re-ingest only when you upload a new version
• Text Sources: Re-ingest only if you’ve updated the content via database

Handling Failures

1. Check the error_message field in the response
2. Fix the underlying issue (e.g., broken URL, corrupted PDF)
3. Wait 1 minute for rate limit to reset
4. Trigger re-ingest again

Automation

Consider automating re-ingestion for URL-based knowledge items:

$
# Example: Daily cron job to refresh all URL-based knowledge items
>
# (Implement your own rate limiting and error handling)
>
for uuid in $(fetch_url_knowledge_item_uuids); do
>
  curl --request POST "https://app.resemble.ai/api/v2/knowledge_items/$uuid/reingest" \
>
    -H "Authorization: Bearer YOUR_API_TOKEN"
>
  sleep 60  # Respect rate limit
>
done

Error Codes

See Also

• Get Knowledge Item - Monitor re-ingestion progress
• Create Knowledge Item - Initial ingestion process
• Update Agent - Configure RAG settings