liveBook · Manning

File upload

As explained in AI Agents in Action

A file upload is an assistant capability for supplying documents or other files as knowledge to a configured language-model assistant.

A file upload gives an assistant static knowledge by adding documents or other supported files to its knowledge base. Detailed RAG knowledge is optional. section 3.4

Configure a document-aware assistant by following these practical steps:

  • Create an assistant in the ChatGPT interface, add instructions, and give it a name and image. section 3.4.2
  • Upload one or more documents; a single assistant supports up to 512 MB, with batch limits varying by file size. section 3.4.2
  • Use prompts to search, compare, contrast, order, classify, or generate content from the uploaded collection. section 3.4.2
  • Warn users about heavy file-upload usage because resource limits can temporarily block accounts. section 3.5.1

Classic Robot Reads

Upload a collection of classic robot books and instruct the assistant to reference only those files, provide three examples for each query, and ask whether more help is needed. Prompts can include “Search for this phrase” or “Identify the three most similar books.” section 3.4.2

Table 3.1 Use cases for a knowledge assistant
Use case
Example prompt
Results
Search
Search for this phrase in your knowledge: “the robot servant.”
Returns the document and an excerpt
Compare
Identify the three most similar books that share the same writing style.
Returns the three most similar documents
Contrast
Identify the three most different books.
Returns books in the collection that are the most different
Ordering
What order should I read the books?
Returns an ordered progression of books
Classification
Which of these books is the most modern?
Classifies documents
Generation
Generate a fictional paragraph that mimics your knowledge of the robot servant.
Generates new content based on its knowledge base

Browse concrete prompts and results for searching, comparing, contrasting, ordering, classifying, and generating from uploaded documents. section 3.4.2

How AI Agents in Action covers it

How deep it goes
Explains setup, capacity, prompts, and usage risks, while deferring detailed RAG implementation.
Examples
Uses classic robot books, phrase search, document comparison, reading-order prompts, and knowledge-constrained generation.
What it assumes
Assumes basic assistant use, teaches file-based knowledge setup, and treats detailed RAG understanding as optional.
What you can do after
Operate/configure a document-aware assistant by uploading files, defining boundaries, and prompting its knowledge base.
How it connects
Connects file uploads with retrieval augmented generation, knowledge assistants, document search, and custom actions.

Also explained in

Compare how 5 books explain file upload →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of AI Agents in Action. The section links point to the parts of the book it draws on.