> ## Documentation Index
> Fetch the complete documentation index at: https://docs.platform.qubrid.com/llms.txt
> Use this file to discover all available pages before exploring further.

# RAG

> Custom multimodal RAG pipeline combines advanced search with the best models to turn your documents, images, and audio into beautifully accurate, cited answers - just upload and we handle the rest.

RAG is your intelligent AI research assistant that combines powerful language models with real-time document retrieval to give context-aware, grounded answers. Imagine chatting with an expert who not only understands your question but also instantly pulls precise information from your documents - that’s RAG. It’s privacy-first, lightning-fast, beautifully formatted, and model flexible - ideal for exploring and understanding your content interactively.

<Note>
  RAG is still in **BETA**, if you encounter issues let us know. Its currently free for users to try
</Note>

## How to use RAG?

<Steps>
  <Step title="Head over to the RAG tab from the left menu">
    This will open up the 2 Step RAG UI, ready to be used
  </Step>

  <Step title="Select the Files">
    You can either click to select or even drag & drop files in the window to get started

    <Info>
      File types: .pdf, .txt, .doc, .docx, .md, .png, .jpg, .jpeg, .mp3, .wav are supported & Total size of all files uploaded can be upto 16 MB
    </Info>
  </Step>

  <Step title="Click the Upload button">
    This will open up the 2 Step RAG UI, ready to be used
  </Step>

  <Step title="Head over to the Chat UI">
    Now you can ask whatever questions you might have, the answers will be from the reference documents uploaded only
  </Step>
</Steps>

<Info>
  If you want to delete the files, simply press the ❌ button beside the documents. To clear the chat & start a new conversation you can use the `Clear Chat` button
</Info>

<Note>
  The answers provided by the RAG Agent will also have citations & will mention the PDF source for you to crosscheck if needed
</Note>

## Why Use Multimodal RAG?

* **Accuracy Across Formats**: Reduces hallucinations by grounding models in trusted enterprise data, not just text.
* **Custom Context**: Aligns responses with your organization’s knowledge base, images, videos, or audio transcripts.
* **Freshness**: Works with frequently updated multimodal datasets (e.g., meeting recordings, surveillance images, reports).
* **Multimodal Use Cases**: Go beyond text-only AI; integrate visual, audio, and structured data reasoning.

## Supported Modalities

* **Text**: Documents, FAQs, manuals, knowledge bases.
* **Images**: Diagrams, charts, product images.
* **PDFs**: Reports, invoices, academic papers.
* **Audio**: Meeting recordings, call center conversations.
* **Video**: Training videos, tutorials, surveillance streams. `Coming Soon`
* **Structured Data**: CSVs, relational tables, analytics exports.  `Coming Soon`

## Example Scenarios

* **Customer Support**: Retrieve text FAQs + product images to answer queries.
* **Compliance & Legal**: Retrieve legal PDFs + annotated charts for audits.
* **Healthcare**: Ground AI in **radiology images + doctor’s notes**.
* **Research**: Combine **academic papers (PDFs)** with **charts/images** from experiments.
* **Enterprise Training**: Use **video transcripts + slides** to answer employee questions.