Document Q&A is a system that answers natural-language questions using the contents of your documents, such as forms, reports or scanned pages.
In plain words
There are two families. Some models look at a document image directly and use its words, their positions on the page and the picture itself, which helps with forms and tables. Other systems work on plain text: they index your documents, fetch the ones most related to a question, and have a language model write the answer from them. This second pattern is retrieval-augmented generation.
A quick example
Hugging Face's task page shows a scanned business document paired with a question about the purpose of a team named on it. The model reads the page and answers with a short phrase about balancing cost efficiency and quality customer service.
Where you'll meet it
- Pulling names, totals and dates out of forms.
- Answering questions over your own document collection with RAG.
Related
Sources
- DocsDocument Question Answering, Hugging Face Tasks · read 29 Sept 2026
- DocsCreating Your First QA Pipeline with Retrieval-Augmentation, Haystack (deepset) · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.