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

# Document Upload — Train AI on PDFs, Word, PPTX

> Train your InsiteChat AI chatbot on PDFs (including scanned PDFs via OCR), Word docs, PowerPoint, Excel, CSV, Markdown, and plain text. Drag-and-drop in the dashboard.

**InsiteChat document upload** lets you train your AI chatbot on files from your existing content library — PDFs (with OCR for scanned documents), Word documents, PowerPoint decks, Excel sheets, CSVs, Markdown, and plain text. InsiteChat extracts the text and indexes it alongside any other sources you've added — website crawls, [Google Drive](/integrations/google-drive), [Notion](/integrations/notion), and [Dropbox](/integrations/dropbox).

## Supported file formats

| Format             | Extensions              |
| ------------------ | ----------------------- |
| PDF                | `.pdf`                  |
| Word document      | `.doc`, `.docx`         |
| PowerPoint         | `.ppt`, `.pptx`         |
| Plain text         | `.txt`                  |
| Markdown           | `.md`                   |
| CSV                | `.csv`                  |
| YouTube transcript | Paste a YouTube URL     |
| Zendesk articles   | Connect via integration |

## Upload a document

<Steps>
  <Step title="Open your dashboard">
    Log in to InsiteChat and select the chatbot you want to train.
  </Step>

  <Step title="Go to sources">
    Click the **Sources** tab, then click **Add source**.
  </Step>

  <Step title="Choose file upload">
    Select **File upload** from the list of source types.
  </Step>

  <Step title="Select your file">
    Click **Choose file** and select the document from your computer.
  </Step>

  <Step title="Confirm the upload">
    Click **Upload**. InsiteChat processes the file and indexes its content. The source card shows a progress indicator and changes to **Ready** when indexing is complete.
  </Step>
</Steps>

<Tip>
  Documents with clear, descriptive headings produce better chatbot answers. InsiteChat uses headings to understand document structure and organize relevant content.
</Tip>

## Best practices for document quality

The quality of your chatbot's answers depends heavily on the quality of the documents you upload. Follow these guidelines to get the best results:

* **Use clear headings** — Structure documents with descriptive H1, H2, and H3 headings to help InsiteChat organize content effectively.
* **Avoid scanned images without OCR** — InsiteChat extracts text from documents. Scanned PDFs containing image-based text cannot be indexed; use selectable text or run OCR first.
* **Keep formatting clean** — Complex tables and multi-column layouts reduce extraction accuracy. Simplify or convert to plain text/Markdown before uploading.
* **One topic per document** — Focused documents are easier for AI analysis. Split large multi-topic documents into separate files.
* **Check for accuracy** — Remove outdated information, correct errors, and delete obsolete sections before uploading.

## File size limits

InsiteChat supports most standard document sizes. For unusually large files, consider splitting them into smaller sections to ensure reliable processing. Specific limits appear in the upload interface.

## YouTube transcripts

To index a YouTube video, select **YouTube** as the source type and paste the video URL. InsiteChat retrieves the transcript and indexes the spoken content for videos with captions.

## Zendesk articles

Connect your Zendesk account from the **Sources** tab to import help center articles directly. InsiteChat indexes article content immediately after import.