Ingest Content | PrivateGPT

Unified endpoint to ingest files, text, URIs, or already processed content

This endpoint expects an object.

artifactstringRequired=1 character

Unique identifier for the text being ingested within the collection

inputobjectRequired

Raw input data to be processed and ingested into the system. Can be a file (base64), URI, or plain text

collectionstringOptional=1 characterDefaults to pgpt_collection

Collection name to group related documents for better organization and filtering

metadatamap from strings to anyOptional

Optional metadata dictionary containing additional document information. If provided, must include ‘file_name’ with a valid file extension

This endpoint returns an object.

datalist of objects

List of ingested documents with their metadata and processing information

object"list"

Response object type, always 'list' for ingestion responses

model"private-gpt"

Model identifier, always ‘private-gpt’

401

Unauthorized

422

Unprocessable Entity

$	curl -X POST https://host.com/v1/artifacts/ingest \
>	-H "Content-Type: application/json" \
>	-d '{
>	"artifact": "company_profile",
>	"input": {
>	"type": "text",
>	"value": "Our company was founded in 2020..."
>	},
>	"collection": "corporate_docs",
>	"metadata": {
>	"file_name": "company_profile.txt",
>	"author": "Marketing Team"
>	}
>	}'

1	{
2	"data": [
3	{
4	"artifact": "company_profile",
5	"object": "ingest.document",
6	"doc_metadata": {
7	"file_name": "company_profile.txt",
8	"author": "Marketing Team"
9	}
10	}
11	],
12	"object": "list",
13	"model": "private-gpt"
14	}