Skip to main content

Library Upload

Your project library is an important source of data for your workflows and implementations. Especially if you are building workflows to interpret large amounts of text and unstructured data. Uploading a quality project library is often the first stage of starting a new project.

When a document is uploaded into Intellectible, a series of processes take place that allow you to use your source material within the platform.

  • First, the text is parsed from the document and converted into markdown. Intellectible is equipped with state-of-the-art parsing engines that allow you to parse and store PDF, DOCX, CSV, text, and webpages.

  • Once parsed, we intelligently split the document text into smaller chunks, making a best effort to retain meaning and structure.

  • Each chunk of text then goes through a complex indexing process, allowing your content to be searchable within workflows and agents.

Library upload process:​

library upload

Disable indexing

It is possible to disable indexing for a document on library upload if you do not need it to be searchable. When uploading the document simply check prevent indexing. Indexing is disabled by default on the Write to Library node but it can be enabled if you wish to index your generated content.

Accessing your library in workflows​

Once your library is in place you can start building workflows that use the content you have uploaded and indexed. For example, you can build custom RAG pipelines that use your library as a source of data for LLMs, for building bespoke agents and custom searches.

Reading parsed documents​

Once there is content in your library, you can access it in workflow in a number of ways.

To read the parsed text from your library, you can use the Read nodes.

These nodes allow you to read the parsed content directly from your library and use it in your workflow.

Searching for indexed documents​

To search for documents in your library, you can use the Search node. The node will return the chunks best matched to your search query based on the library or a subset of the library you have selected.