RagieAI Components

Creates a new document in RagieAI from a URL.

Ingest a document as raw text. The document goes through processing steps (pending → partitioning → partitioned → refined → chunked → indexed → summary_indexed → keyword_indexed → ready → failed). The document is available for retrieval once it reaches the ready state. The summary index step can take a few seconds. You can optionally use the document for retrieval once it is in indexed state, but the summary will only be available once the state has changed to summary_indexed or ready.

Deletes a document from RagieAI by ID.

Lists and searches documents in RagieAI. Returns a maximum of 100 documents per page.

Retrieves a document from RagieAI by ID.

Get the content of a document. The mediaType parameter can be used to request the content in a different format. When requesting as application/json, additional metadata about the document will be included. If the original document contained content such as images or other non-textual media, this response will include a text description of that media instead of the original file data. Using mime types such as audio/mpeg or video/mp4 will stream the file in a format that can be provided to an audio/video player.

Update metadata on a document. Performs a partial update of the document's metadata. Keys must be strings. Values may be strings, numbers, booleans, or lists of strings. Keys set to null are deleted. Maximum 1000 total values allowed, inclusive of existing metadata.

Updates a document from a publicly accessible URL. The document goes through processing steps (pending, partitioning, partitioned, refined, chunked, indexed, summary_indexed, keyword_indexed, ready, failed) before it is ready for retrieval. PDF files over 2000 pages are not supported in hi_res mode.

Update a document with raw data in text or JSON format.

Performs an arbitrary authorized API call to the Ragie AI API.

Get a LLM generated summary of the document. The summary is created when the document is first created or updated. Documents of types ['xls', 'xlsx', 'csv', 'json'] are not supported for summarization. Documents greater than 1M in token length are not supported. This feature is in beta and may change in the future.