AI RAG API

Endpoints for managing vector storage used in Retrieval-Augmented Generation (RAG). These endpoints allow you to create vector stores, upload content for indexing, and manage stored data.

SDK equivalent: sdk.ai.rag

Create Vector Storage

Create a new vector storage instance for a specific scope.

POST /api/ai/sources/{sourceId}/storage

Path Parameters

Parameter	Type	Description
sourceId	string	The AI source ID

Request Body

{
  "integrationId": "integration-id",
  "scope": "thread",
  "subjectId": "thread-id",
  "expirationAfterDays": 30
}

Field	Type	Required	Description
integrationId	string	Yes	The AI integration ID
scope	string	Yes	Storage scope: thread, user, workspace, or tenant
subjectId	string	No	The subject ID (e.g., thread ID, user ID)
expirationAfterDays	number	No	Auto-expire storage after N days

Response

{
  "success": true,
  "message": "Vector store created",
  "vectorStore": {
    "_id": "vector-store-id",
    "scope": "thread",
    "subjectId": "thread-id",
    "externalId": "external-id",
    "expirationAfterDays": 30,
    "created": "2025-01-01T00:00:00.000Z"
  }
}

SDK: sdk.ai.rag.createStorage(sourceId, data)

---

Upload Content

Upload content to be indexed in a vector store for RAG retrieval.

POST /api/ai/sources/{sourceId}/storage/upload

Path Parameters

Parameter	Type	Description
sourceId	string	The AI source ID

Request Body

{
  "integrationId": "integration-id",
  "content": "This is the documentation for our product...",
  "fileName": "product-docs.txt",
  "vectorStoreId": "vector-store-id",
  "metadata": {
    "type": "documentation",
    "version": "1.0"
  }
}

Field	Type	Required	Description
integrationId	string	No	The AI integration ID
content	string | object	Yes	Text or JSON content to index
fileName	string	No	Name for the uploaded content
vectorStoreId	string	No	Target vector store ID
metadata	object	No	Additional metadata for the content

Response

{
  "success": true,
  "message": "Content uploaded",
  "fileId": "file-id",
  "vectorStoreId": "vector-store-id"
}

SDK: sdk.ai.rag.uploadContent(sourceId, data)

---

Clear Storage

Remove files from a vector store.

POST /api/ai/sources/{sourceId}/storage/clear

Path Parameters

Parameter	Type	Description
sourceId	string	The AI source ID

Request Body

{
  "integrationId": "integration-id",
  "vectorStoreId": "vector-store-id",
  "fileIds": ["file-id-1", "file-id-2"]
}

Field	Type	Required	Description
integrationId	string	No	The AI integration ID
vectorStoreId	string	No	The vector store to clear
fileIds	string[]	No	Specific file IDs to remove. If omitted, clears all files.

Response

{
  "success": true,
  "message": "Storage cleared",
  "clearedCount": 2,
  "vectorStoreId": "vector-store-id"
}

SDK: sdk.ai.rag.clearStorage(sourceId, data)