Skip to main content
Upload your files to the Data Engine so they can be searched and used to ground AI answers. Gloo AI pulls out the text, splits it into searchable pieces and stores them so they can be found by meaning. Within minutes, your content is ready for search and AI answers, with no preprocessing or custom pipelines. Building your own upload, parsing and indexing pipeline can take weeks; the Files API handles it for you, with automatic parsing and indexing for all major file types and no manual steps, availability in your Gloo AI workspace as soon as processing finishes, and performance that scales to millions of files. You can upload with the API or from Gloo AI Studio without writing code.
You must have at least one Publisher in your organization before uploading. See Manage Publishers.

Prerequisites

You need:

Upload Files

Send files to the ingestion endpoint. You can mix file types freely; the endpoint detects and processes each format.Every upload must include a publisher_id form field identifying the Publisher the content belongs to. The files field is also required.
Example Response:
You can upload multiple files in a single request by adding multiple -F "files=@..." parameters.

Upload with Your Internal ID

Use the producer_id query parameter to associate your internal ID with the uploaded file, so you can look up items later by your own identifiers. For guidance on choosing a producer_id scheme that stays stable across renames and syncs, see Preparing Content for Ingestion.
Shell
A multi-file upload that includes producer_id is rejected with a 400 too_many_files error. Producer IDs must have a one-to-one relationship with a file.
Gloo doesn’t replace items by producer_id. Uploading changed content under a producer_id you’ve already used creates a second item; to change an existing item’s content, replace it as described below.

Replace an Item’s Content

Gloo detects duplicates by file content. Uploading a file whose bytes are identical to one already in the publisher’s library doesn’t create a copy: the existing item comes back in the response’s duplicates list. To change what an existing item contains, upload the new file with the update query parameter set to that item’s item_id:
Shell
Gloo deletes the existing item and processes the new file under the same item_id. Send one file per request, and send the item’s producer_id again so the replacement keeps it. Metadata you set on the old item isn’t carried over, so re-apply it with Edit Content. If you only have your own ID, find the item_id with Look Up by Producer ID. If several items share that producer_id, the lookup returns the newest one, and replacing it leaves the older copies in place: delete them, or keep the item_id you mean to replace yourself. The existing item is deleted before the new file is processed. If the new file is identical to a different item in the publisher, the response lists that item in duplicates and no item is left under the item_id you sent.

Set Chunk Size and Overlap

Add chunk_strategy, chunk_size or chunk_overlap to an upload to change how a document is chunked. For the allowed values and defaults, see Set Chunk Size and Overlap.

Next steps

To update metadata after upload, see Edit Content. To learn why content is chunked and indexed, see Building a Knowledge Base in the AI Learning Center. For every request field, see the API Reference. To check a file’s ingestion status by its item_id, use Get File by Item ID.

Search what you uploaded

Find passages in your content by meaning.

How Chunking Works

Understand what happens to your files after upload.

Manage Content

View and manage your content, including looking up items by producer ID.

Upload Files Recipe

More code examples for uploading files.